The Linux desktop has come a long way. It’s a fully usable, stable, and secure operating system that can be used quite easily by the masses. Not too long ago, Sun figured they could do the same by starting Project Indiana, which is supposed to deliver a complete distribution of OpenSolaris in a manner similar to GNU/Linux. After using the latest version for a while, I’m wondering: why?
I was very interested in trying out the latest OpenSolaris release, hoping to find an operating system that could rival the likes of Ubuntu and OpenSUSE. The goal of Project Indiana is basically to create GNU/OpenSolaris, slap on a GNOME desktop, and try to capitalise on OpenSolaris’ key features, among which ZFS. I had the plan to write a nice review of it, but the experience was so bad, I had to let go of that plan.
The experience was nothing short of a major disappointment. I think if there is one word that really summarises the entire experience, it would be “slow”. Installation is slow. Booting takes forever. It takes far too long for applications to load or for windows to appear. Responsiveness is terrible. On a slightly less important note, Compiz performs much worse on OpenSolaris than it does on Linux. On top of that, the package manager is slow during all operations – even downloading the packages. It was a horrible experience. I was using it on a 2.8Ghz Pentium IV with HyperThreading, 2GB of RAM, and an NVIDIA 6200 video card.
Of course, the OpenSolaris effort is still relatively new, so we should cut the project some slack. They should be given the time to mature the project, and work on improving the overall performance. In addition, they should also work on getting more packages in the repositories, because I could just about make out the tumbleweeds rolling across my desktop.
However, even if they did improve the overall experience (again, especially overall performance and responsiveness), there’s an additional problem that I think is a little harder – if not impossible – to solve. While I was using OpenSolaris, it dawned on me that if it weren’t for the custom theme, there’s very, very little to set OpenSolaris (the Project Indiana variant, that is) apart from various Linux distributions.
And distributions like OpenSUSE, Fedora, and Ubuntu all provide a far, far better user experience than OpenSolaris does. While I see no problem in Sun trying to turn OpenSolaris into a Linux distribution (effectively) I’m left wondering if that’s enough. I have the idea that if Sun wants OpenSolaris to achieve the same kind of success that desktop Linux distributions have seen, they’ll have to offer something different.
As it stands now, OpenSolaris felt like a really, really bad Linux distribution from 5 years ago. I’m not saying that Solaris is a bad operating system – oh no – I’m just saying that it’s a bad Linux distribution.
What are your experiences with OpenSolaris?
I pretty much came to the same conclusions that you did the few times that I’ve gone and loaded up OpenSolaris or even Nexenta in VirtualBox, never had the cajones to actually install it to disk and see how much fun I have with that. The install process is painfully slow I noticed and the actual user experience is just so incredibly lacking. I like some of the look and feel in OpenSolaris but think it needs some TLC in a few areas, it just kinda feels bland and, I dunno, corporate? Application availability is understandably sparse as far as the repositories go, and that’s something which is hard to fault the developers on because really, if you wanna have a respository as big as Debian, et al. then you need to have as team as big as they have. I can easily see OpenSolaris catching on with a certain crowd because it offers a neat feature set, though for a decidedly niche market at least for now. Your average desktop user does not need nor want something like ZFS, at least not until it becomes a little less resource hungry though I have no idea how they will manage that due to the inherent design of ZFS. Half a gig to 1 gig of memory for the filesystem is really asking quite a lot still of the average user. Dtrace is supposed to be great for developers but again, your average desktop user has absolutely zero need or desire for this.
Ultimately though driver support combined with a lack of performance optimization are the biggest barriers here. I didn’t speak much about Nexenta by the way because when I tried it in my virtual machine I gave up after discovering the default install didn’t put X and GNOME on, and the amount of effort it took for me to get just the desktop running was a big turn off for me.
Even with all of this said, though, I probably wouldn’t mind any of the systems shortcomings as long as I could play my games. Wine is supposed to be working on OpenSolaris well enough to play World of Warcraft, for example. http://opensolaris.org/jive/thread.jspa?threadID=33766&tstart=0 Now if only all my other games would run so happily under Wine…
Your average desktop user would absolutely love something like ZFS with all the snapshot-y goodness, especially when it is integrated into something like TimeSlider in Nautilus. It’s similar to how TimeMachine works on MacOS X, but with better technology (ZFS) behind the scenes. If Apple ever gets around to completing ZFS support in MacOS X, they’ll have a truly killer feature once TimeMachine makes use of it.
Absolute FUD. You can run ZFS on 32-bit systems with as little as 512 MB of RAM (total system RAM). You can also run in it on 64-bit systems with 64 GB of RAM. And everything in between. ZFS works better with more RAM, and can do more caching as RAM increases, but it can be tuned to run in very low memory setups.
When the average desktop user begins to understand even something like System Restore then I will concede that the average desktop user might appreciate something like automated snapshots. A small amount of people make a large amount of fuss over snapshots which I argue A) most average desktop users are not technical enough to understand or utilize properly and B) waste a lot of disk space when enabled by default and utilized by few. Maybe I’m just out in the dark here because nobody I talk to outside of tech enthusiasts or people creating an infrastructure that can utilize these features really care about this particular feature. It is wonderfully useful for certain environments, but I don’t think the average desktop user needs this kind of functionality. I would turn it off because I frankly don’t need the overhead and have never thought to myself “Gee, I really wish I could revert this file back to a previous version, or undelete something I still want after all”. Most of my unintentional data-loss is because of things like a partition getting FUBAR’d, not because one way or another the files ended up deleted.
What I said was not FUD, you explicitly confirmed what I had said which was “Half a gig to 1 gig of memory for the filesystem”. 512MB = Half a gig? I realize that ZFS works better with more RAM, but my point is that I don’t see its feature-set as an acceptable trade-off on your average desktop.
ZFS is a great filesystem for a lot of purposes, but I just don’t see how right now anyone could seriously consider any of its features as something vital the average desktop user needs to be exposed to. Ugh, have said “average desktop user” way too many times already and just did again.
Checksumming of your data is vital. Ever had data corruption? Is your data still ok? how do you know there is no silent corruption? In that case you really want to know which files are broken and which are not.
Compression, pooled storage, snapshots and encryption(?) are nice, but not a must have for the average poweruser.
All of these features are great things that I would love to toy around with, but you said it, not even your average power user is going to make much of a deal about them. Will be very nice once btrfs matures a little bit more and I can play with such features more easily in an environment that I am comfortable with.
Checksumming of your data is vital. Ever had data corruption? Is your data still ok? how do you know there is no silent corruption? In that case you really want to know which files are broken and which are not.
No, checksumming is not “vital”, specially considering that self-healing, which is the big deal about checksumming and detecting corruption, requires to downgrade the size of your hard drive to the half, and desktop users won’t accept that. The fact is, most of desktop people (which these days upload big portions of their personal data to the “cloud”) don’t suffer data corruption. The few desktop users that have data corruption many times can afford losing the data (in my experience), and for rest of the people that loses important data there’s always a lesson to learn: backups. (remember: checksumming and self-healing are just a great way to avoid the need of a backup restore, but not a backup)
The fact is that hard drives are not the unreliable thing that you seem to want to think. Most of people needs to buy a new computer before they hit data corruption. If drives aren’t safe people won’t buy them, and the market will choose the drives from the best company. So most of the people just doesn’t have problems with that. Obviously, it’s a great and much needed addon, but a requirement?
Why would you have to cut your storage space in half in order to gain the benefits of self-healing and error detection? Or are you talking about single harddrive setups?
Why would you have to cut your storage space in half in order to gain the benefits of self-healing and error detection? Or are you talking about single harddrive setups?
Single or multiple, what matters? You need to cut your storage pool space to the half to duplicate all the data needed to self-heal…(not for metadata, obviously, since metadata gets duplicated 2/3 times even on a single disk with no data replication). Good luck convincing desktop users that they need to lose all that space…
Edited 2009-07-20 22:22 UTC
If you have a single harddrive, then, yes, you will have to halve (or more) your storage, by setting copies=2. Without that, you don’t have redundant storage to compare against.
But on multiple-harddrive systems, you configure the storage pool using mirror or raidz vdevs. Since pretty much everyone already does this using RAID, whether it be hardware or software, you’re not “losing” anything.
How do you expect to have self-healing, without redundant storage? And why would you consider using redundant storage to be bad?
And why would you consider using redundant storage to be bad?
Again: Try convincing desktop users that they need to buy extra storage, or cut the half of their storage (even when they buy multiple hard drives)
In large multiple drive systems you mostly end up with far more redundancy with ZFS than you actually practically need. It’s one of the reasons why Hammer came about.
System Restore on Windows is *nothing* like TimeMachine or TimeSlider or even plain ZFS snapshots. And the UI for System Restore makes it even less similar (is it just me or does MS go out of its way to make simple things/concept hard to use?).
ZFS snapshots use 0 diskspace until something changes. So if none of your files change between snapshot A and snapshot B, then 0 disk space will be used. If only 1 file changed, then only the space for that one file will be used. And so on.
I use ZFS on my home media server, taking automatic snapshots every evening, and I keep 60 days worth of snapshots. Total snapshot disk usage is ~5 GB, for 250 GB of total disk space in the server. Several times now I’ve used the snapshots to recover files accidentally deleted by myself or the wife. And to recover older versions of files (like resumes and business letters).
(Un)fortunately, my system runs FreeBSD, so the nice TimeSlider GUI isn’t available (not that I’d use it, since I can’t stand Nautilus). But even using the shell, it’s simple to “recover” files from snapshots (cd /path/to/filesystem/.zfs/snapshot/snap-name/; cp /path/to/wherever).
Talk to MacOS X users who use TimeMachine. You’ll hear a different side of things (“OMG, how’d I ever live without this?” is the common response I get).
No, you’re failing to see the difference between “512 MB of RAM for the filesystem” and “512 MB of RAM for the whole OS, including the filesystem”.
You don’t need 512 MB of RAM *just* for ZFS. You can use ZFS on a system with only 512 MB of RAM. Very big difference.
IOW, if you have a lowly laptop or desktop with only 512 MB of RAM, total, you can still run a system using ZFS as the main filesystem.
They can serve the same purpose (though ZFS and the others can serve far more purposes than simple system recovery) regardless of how they go about doing that, though there is no doubt that System Restore is junk in comparison to Time Machine and similar technologies.
Regarding how many GBs your snapshots take up with your own usage patterns, like you said this kind of stuff all depends on what you do with the system. I can’t imagine that a whole lot changes on a daily basis with your home media server, which is why the ~5GB snapshot disk space usage sounds about right. Just because you only use up ~5GB with snapshots doesn’t mean that’s what another user should expect. Besides, it was never really my intention to talk about the possible scenario of running low on disk space and having multiple GBs of snapshots to get rid of, but rather to address the issue that as far as I’m aware it’s a disk space trade-off for a feature that doesn’t have an incredible amount of practical value for your average user.
If that’s their response then they don’t sound like they’re very responsible users, but considering a lot of audio and video production occurs on Macs it’s not very hard to imagine that a lot of these people appreciate the ability to easily revert to previous versions of things they have been working on. This has little to no practical value to your average desktop user who does not create content, but merely views it. If these people are not professionals or enthusiasts, then I have to ask what they’re doing that makes Time Machine such a deal-killing feature, something they could never go without anymore. Of course system/network admins would probably like these features for slightly different reasons.
These are pure semantics, I thought it was understood that saying a 512 MB of RAM for the filesystem was the same as saying 512 MB of RAM for the whole OS, including the filesystem.
Furthermore, just because you can run OpenSolaris with ZFS on a lowly laptop or desktop with 512 MB of RAM doesn’t make it the best idea. Did you hear the story about Windows 7 being hacked down enough to run on a system with a paltry 96 MB of RAM? Sure you could do it, but it’s a pain in the butt to use when you could really just use something better suited.
To the person who down-voted my comment: Where was the troll, off-topic or inaccurate information in my post? *sigh* It’s ridiculous that people down-vote just because they don’t like what a post is saying, that’s not the purpose of the system.
I said myself that I like OpenSolaris and would probably use it on my desktop with little to no issue, so I can’t be trolling. The subject of discussion is our experiences with OpenSolaris, since Thom asked at the end of the article, so I can’t be off-topic. There’s also nothing inaccurate in my post, so it can’t be that. Oh well.
Utter bullshit.
You cannot run OpenSolaris with ZFS on a 512mb machine and start apps. It just does not work. It nearly does not work with 1GB in my experience. Programms take ages to start it is ridiculous. It is just crap. You just have to have 2GB.
Sorry, but that isn’t true. While it is true that performance will increase with more memory, you can definitely run OpenSolaris on zfs with 512MB of memory. It is the *minimum* requirement for OpenSolaris, and is an officially supported configuration.
It’s the minimum in the same way that Windows XP Professional wants a minimum of 64MB of RAM, stop acting like it’s something anyone would feel comfortable using. I can’t even stand using XP on the machine my sister and her husband have, and that has 192MB of RAM! A running system is pretty severely handicapped if as soon as you actually try to do anything you start hitting swap. VirtualBox will default your OpenSolaris VM to 768MB of RAM for a reason, because it runs like a dog w/less.
Ahhhhhh. The sad ramblings of someone who believes that telling people of the technical merits of something will prove to be good enough. It might prove to be a differentiator when all other things, especially application availability, are equal. Everything else, however, is not equal and Nautilus isn’t even a good enough file manager and is the same as you’ll find elsewhere.
For the benefit of the uneducated I shall translate. You can run ZFS on systems with lower memory requirements, but you will have to tune it if you want it to run trouble-free with acceptable performance. All the evidence thus far says that ZFS grows unbounded to whatever workload you throw at it and you will need to create those boundaries yourself.
The day you see ZFS running on an ARM NAS system with 128MB of memory is the day you see Satan skating to work. They just aren’t going to materialise.
Edited 2009-07-20 23:37 UTC
I disagree with your suggestion that the average user doesn’t care about ZFS. That might be true for the awareness of what their filesystem is actually called (just like ext3, btrfs, …), but they’re definitely aware of the features it provides. Being able to take regular snapshots of their data is *incredibly* useful – if you don’t think so, you’ve obviously not accidently lost some data. At some point in the future we’ll also be able to provide encryption support at the file system level.
IMO, these days, on desktops apps is only what matters. It doesn’t really matter that Solaris (or freebsd) are good or bad kernels (it doesn’t even matter a lot if their filesystems are superfast or just fast, in the surface they are all pretty much POSIX filesystems, only their internal implementations differ), on the surface it’s gnome/kde/xfce + openoffice + firefox + dbus + etc. Even in the server side, what matters is mysql + apache + etc. So yet another gnome/kde distro is not going to be succesful just because it has a different kernel. People doesn’t really care, so they use the standard gnome/kde distros, which happen to be linux-based.
Opensolaris is just a kernel with dtrace and ZFS. Don’t discount the usefulness of either of those, even if you are just running the Lamp stack. (Hint Filesystem performance is sorta really important for databases, as is the scheduler).
But, yes, having dtrace and ZFS(not in FUSE) available in a typical linux distro, would be better.
Well I would argue an operating system is just a kernel + drivers + (consistent) interfaces.
But you write as if that is a bad thing!? Weren’t Microsoft (rightly) lambasted for suggesting a browser was part of their operating system And then a media player. And …
OpenSolaris is certainly more than just a kernel, but the fact/problem is the kernel (+ZFS) is it’s best bit! As Thom found out users care less and less about the kernel and more and more about the flashy effects and familiar desktop applications, which is Opensolaris worst bit!
FWIW on not so different hardware from Thom I can see no noticeable difference in performance between say openSUSE 10.3 and OpenSolaris. 2008.11 (it’s dual boot) But it looks like I spent a bit more time than Thom on it, YMMV.
KM
OpenSolaris is not just a kernel; unlike Linux.
OpenSolaris is a tightly-integrated operating system comprised of a kernel, drivers, system libraries, management frameworks, and other technologies.
For example, fault-management, the dtrace framework, zfs, and more are all examples of technologies that are part of the kernel, but also required changes to applications delivered with the OS to take advantage of them. So no, OpenSolaris not just a kernel.
OpenSolaris is a tightly-integrated operating system comprised of a kernel, drivers, system libraries, management frameworks, and other technologies.
…which is a tiny portion of all the software installed in a common distro, and pretty much all desktop software doesn’t really care about “management frameworks” (except everything about hardware management, which is abstracted properly in HAL/DeviceKit). They care about basic POSIX operations. In fact they don’t even care about that, most of them can run/will run on Win32. The fact that many FOSS apps are so portable shows how irrelevant kernels have become – there is nothing on linux or solaris or any other kernel that “ties” the app to that specific kernel and makes impossible to port the app to other systems.
At the end of the day, what you have is the Firefox/Evolution/Nautilus/Openoffice GUI. Users press buttons and the apps do something. There is nothing special in kernels these days that make the buttons better, and they can’t improve sucky aplications either. The last time I saw kernel changing something on the desktop was with the nautilus “snapshot” functionality that only works in ZFS systems – and that can be emulated with LVM and the corresponding NT equivalent…
Management frameworks, etc. matter more than you want to admit. Management frameworks deal with things like network failures, configuration, etc. which obviously *do* matter to users — as just one example.
Another item is the audio subsystem, Boomer, which unlike Linux-kernel-based OS distributions has fully virtualized audio much like (apologies for the comparison) Windows 7 or Vista. And depending on your hardware, it can support surround sound as well.
So management frameworks and subsystems do matter, and a key differentiator for OpenSolaris is that many applications have been tightly integrated with those management frameworks and subsystems. As opposed to the usual GNU/Linux approach, which is to cobble together a disparate set of software components without tightly integrating them.
All of the lower level stuff like management frameworks and subsystems definitely matter. A bad kernel is going to result in a slow and/or buggy system overall. However, the kernel is not what users see. Your truly average desktop user doesn’t even know what a kernel is. The average linux user might, but even then, what they generally care about is the applications. The kernel and all of the low level packages on a system are things that the user generally just wants to want and not care about. The more techy and geeky a user is, the more they’re going to care about it and whatever nice features it may have, but they low level stuff isn’t really what the user sees and cares about.
So, while the low level stuff is very important, it’s not what users generally care about. What they care about are the desktop applications that they directly use. So, to the average user, management subsystems don’t matter – as long as they work well enough not to cause them any problems anyway.
And if OpenSolaris is using the same desktop and applications as a linux distro, then it’s as good as another linux distro to most people. There are plenty of people who care about things like ZFS and the various cool features in OpenSolaris stuff, but the average user really isn’t going to.
Your point is valid, but does not coutner what I said at all
In particular, my point was that the underlying management framework and kernel can have a drastic effect on applications and the desktop.
For example, the new audio subsystem in builds 117+ provides fully virtualized audio allowing multiple applications to use the sound device at the same time. While this was solved by sound servers, PuleAudio, etc. in the past, OpenSolaris has a fully native implementation. It should be obvious why users indirectly care about this.
Except it isn’t, because OpenSolaris fully integrates additional OS-specific functionality into the “same desktop and applications.” For example, because of zfs, GNOME Nautilus has a time slider feature that is not present on GNU/Linux distributions.
So again, the underlying technology matters when it comes to OpenSolaris. Many of its technologies can have a fundamental effect on applications and how they are used.
Would you mind providing other examples for OpenSolaris features directly included in the GNOME ui, except for Time Slider? This is one of the main complaints I (still?) have to make about it – there’s next to nothing on the desktop which is “unique OpenSolaris”… except for Time Slider, maybe, which also just recently has been included…
K.
If by GNOME UI, you mean GNOME desktop applications, then:
Another example would be the Package Manager, which has support for OpenSolaris’ unique Boot Environments technology.
This allows a user to create and manage boot environments.
Boot environments allow a user to upgrade a “copy” of their currently running system instead of the live system itself preventing the update process from destabilising currently running applications.
If the new boot environment fails (because of some bugs specific to their hardware perhaps), or a new version of their application doesn’t work right, users can simply go back to the old boot environment without losing productivity.
If by GNOME UI, you mean the standard GNOME programs itself?
Here are some examples:
* The GNOME volume control applet has been specifically enhanced to take advantage of and use features found in the new boomer audio subsystem
* In the very near future, the Shutdown dialog for GNOME will be enhanced to allow the user to choose between a ‘Fast Reboot’ which causes the OpenSolaris kernel to reload itself without causing the system to reset (which avoids having to wait for the BIOS POST process, etc.) and a ‘Restart’ which physically restarts the system.
* The GNOME ‘Users And Groups’ application was enhanced to allow assignment of OpenSolaris’ roles and profiles security attributes.
* The network administration control panel of GNOME was enhanced to allow access to OpenSolaris-specific network technologies such as network profiles (even more enhancements are coming in the future).
* A special ‘services’ application was added that allows management of OpenSolaris SMF-based services.
There are probably other areas as well that I’ve missed, but that should give you an idea…
The package manager indeed is unique inasmuch as that it works with the OpenSolaris IPS packaging system. In terms of usability or even speed, however, so far it is not even close to in example synaptic on a Debian derivative. Mainly, it’s way too slow for that.
Yes, the BE feature is pretty cool, even though this is also something I way more want to have on a server on which I have way stricter downtime considerations to do.
Thanks for the pointer, I will have a look at this.
So far however, on OpenSolaris my notebook was quiet at the moment anyhow because, even though the audio drivers listed my machine as “supported”, they somehow simply didn’t work for whichever reasons. Gonna see whether the new builds still show this problem.
Asides this, so far I have no comparison of boomer and the GNU/Linux ALSA architecture in terms of features and/or UI support.
Cool.
Next thing, however, would be to have these things generally available independent of the GNOME desktop. Most of my Unix using developer friends use KDE for good reasons, and I personally use XFCE both because of its overally “lighter” memory footprint and because of nostalgic issues (it reminds me of CDE, simply put… 
). So far, the GNU/Linux distributions more and more provide infrastructure to expose features like this independent of the actual desktop environment used…
Ok. This is a good point.
This is cool but at the moment I just would be happy if I could use the OpenSolaris network connection control to connect my notebook with our WLAN without too much pain. In Xubuntu, this is a matter of having all the data at hand and spending a minute using nm-applet or wicd or whatever. On OpenSolaris, the last time it took me half a day and by now all I came up with was a half-baked scripted solution not working out of the box in a sane way. Didn’t even try doing UMTS dialup via bluetooth, so far…
Not too different to the GNOME service manager on other platforms however.
Yes it does, thanks!
Don’t underestimate the System. Solaris is, yes, just another POSIX OS with the same X, the same GNOME, etc.. But, where Linux is schizophrenic, Solaris is not. This may seem like a minor detail, but it results in a sometimes very different system. FreeBSD is similar for a lot of the same reasons; though it is different from Solaris it is similarly dissimilar from Linux, because it isn’t schizophrenic and has a coherent design from bottom up through… not the top, but close to it.
So, OpenSolaris may at first blush seem just like another Linux distro, but the different details of its underpinnings allow it to be sometimes saner, sometimes easier and sometimes better. And, sometimes, it’s worse.
There is very little in Solaris that could not be cloned by Linux, even not counting specific technologies like dtrace. What can’t be cloned is the process, the control, and the cohesiveness.
I’ve tried on several laptops lately, but was unable to get everything working. My first experience (I think it would be 2008.10 ??) was funny. Audio, video, sound all worked, but couldn’t get the touchpad or the eraser-head (IBM) to work, no matter which I enabled (including both) in the bios. My last try was on a netbook (Asus 1000HE) with the latest build, but could get no networking – kind of a problem on a netbook! I wasn’t able to use it long enough to assess performance, but I’m still going to keep trying every so often. Just for the geek factor, but I imagine I will probably switch back.
It reminds me of the early days of Linux. I would try it on my laptop, but could never get everything working so I would go back to Windows. Now Linux runs fine on my laptops – I imagine OpenSolaris could too, but I doubt they will have the resources to do so unless Oracle goes Solaris crazy. Maybe Larry could do a Ballmer dance for Solaris!
I’d encourage everyone trying it on their laptops to continually (or periodically) check out the latest releases from opensolaris.com – they do include significant amounts of changes to hardware drivers. If you do successfully install it, you can also try following the developer stream (which can at times be unstable), but changing you default repository to pkg.opensolaris.org/dev
I’ve not seriously tested OpenSolaris, but for me it has installed just fine on a small range of hardware, including an old P4 based HP machine with Intel GMA something or other for graphics and a whopping 2 gigs of mem. The only one item it didn’t have a driver for was the win-modem in the HP, and the system felt fairly snappy.
Lack of familiarity coupled with lack of man pages the last time I played with it are the only things that really kept me from sticking with it. More recently I’ve been wondering if Oracle is going to be keeping the project around at all so I guess it’s not the best time to get too attached to the system anyway.
It would be very useful if you could post some bugs on what issues you encountered (http://defect.opensolaris.org). Usually man page support is excellent, though there have been some issues with providing re-distributable manpages – some of which have been encumbered in the past due to 3rd party licensing. A lot of those issues have been resolved in recent releases though.
You can have slickest, most secure kernel with demonic bit crunching strength but if your OS doesn’t have software that can competite with other OS’s software and offer something more, your kernel isn’t going to move further than on the desks of few hobbyists.
Truth… Slowlaris got its name for a reason.
It was reliable in production, but it was never ‘fast’…HPUX usually smoked it in performance.
Most people have dumped them both and their brothers (AIX, etc) for Linux at this point anyway.
Running solaris is like running a copy of VMS…its useful for historical purposes only.
You’ve got a point, though AIX is still doing pretty strongly in the banking space. Considering how the banking space is doing though, that doesn’t really mean much at the moment. 😉
I always found Solaris a strange beast. I’ve only ever seen it outperform AIX once in the DB arena but that was with a strange custom app with db/flatfile hybrid backend (!?!). Obviously the app was highly optimized for Solaris on SPARC but it does prove that it can be done.
As for Linux adoption, it’s currently sky rocketing among former Solaris houses. I’ve been speaking to some Sun engineers here in Luxembourg and they’ve been telling me all kinds of horror stories of people jumping ship ever since the company has been up for sale. So many of them are worried about losing their jobs, it’s really not a pretty sight.
I guess we’ll have to wait and see if the acquisition by Oracle will actually chnage any of that.
No doubt that future acquisitions are causing uncertainty, but a lot of judgements around performance etc have been made on 2 year+ old operating systems – I’ve seen a lot of FUD recently about how someone should migrate to Linux because they’re doing a comparison with Solaris 8 and Solaris 9. A lot of enterprise systems will obviously install once and leave for several years, but to ignore Solaris 10 would be foolish IMO.
If OpenSolaris is a Linux distribution, then isn’t it missing…um…Linux?
As long as you’re talking about Linux-the-kernel, yes.
Now, let’s talk about Linux-the-distro. It’s usually Linux, glibc, X11, {kde,gnome,xfce,lxde}, more apps, some random packaging horror and a custom theme to make you feel at home at the developer HQ.
OpenSolaris replaces Linux and glibc with its own counterparts, but the other stuff is there. From a user perspective (Kernel and libc are really not that interesting for users), it’s the same.
I honestly don’t know why people bother trying opensolaris. Its crap. Solaris Express Community Edition (SXCE) is where its at. OpenSolaris is based on SXCE, but with the IPS package manager. I never really liked IPS when I tried opensolaris. However, it will take a little more effort to get SXCE into a usable desktop form.
There are always trolls bitter about the change in packaging systems, but in short, the new package system is still light years better than the old one in terms of usability.
The old SVR4 packaging system was positively ancient compared to almost any other modern packaging system: no unicode support, no remote search capability, primitive publication tools, no dependency analysis or constraint capabilities, required downloading entire package for upgrades instead of only changes, arbitrary limits on lengths of package names and descriptions, and was not designed to accommodate third-party network repository usage.
SXCE is OpenSolaris plus a bunch of useful nonfree packages.
It’s not a troll to ask why they didn’t just use the Debian package system (or rpm or) instead of rolling their own and leaving their would-be users with very few packages and new buggy tools to annoy.
And that Ian was involved in this … it hurts my brain.
“It’s not a troll to ask why they didn’t just use the Debian package system (or rpm or) instead of rolling their own and leaving their would-be users with very few packages and new buggy tools to annoy.”
A lot of people have asked this question, and a lot of it has been around the fact that to do package management on OpenSolaris you have to worry about a lot of other things that are substantially different compared to Linux – SMF, various virtualization technologies, ZFS, etc… The thought was the number of patches it would take to port dpkg/rpm/whatever across would be pretty substantial with a strong likelihood of having to maintain them since they wouldn’t get merged upstream. IPS is evolving technology and does have issues, but it’s getting better and better with every release (we hope to solve the package renaming issues to move away from confusing SUNW prefixes with the next release btw)
SXCE will likely go away at some point in the future. What you are seeing with OpenSolaris will form the basis of a future generation Solaris. IPS is something that is continually evolving. What do you hate about it?
I’m curious as to what you define as ‘slow’ installation. For example, on three different systems I’ve tried installation completes in fifteen minutes or less.
If you have a time significantly longer than that, then there’s a good chance you’ve encountered a bug. However, without more information it is difficult to assess your installation issue.
OpenSolaris is primarily targeted at developers, so if you’re not a developer, I could also see why you don’t believe that it doesn’t offer anything beyond many GNU/Linux distributions.
However, if you’re a developer, it offers a lot more:
* a free, professionally supported compiler suite (Sun Studio)
* DTrace
* complete documentation for system programming interfaces
* stable kernel API and ABI (drivers and programs written twenty years ago or more will still work on many systems)
With that said, OpenSolaris does offer a few unique things that GNU/Linux distributions don’t have to users that aren’t developers: time slider, low-overhead quasi-virtualization with zones, xen support, and starting with build 117 (the /dev release of OpenSolaris) a sound system that natively supports surround sound and virtual audio mixing (and doesn’t rely on user-space hacks like pulseaudio).
As for the package management system, it’s still under heavy development and design. There are many improvements still needed, but I think with time it will offer some unique functionality not offered by other systems (some of which can already be seen).
I used Project Indiana from the initial release and had laptop wifi access to my network and everything else fell into place. Since then, it’s gotten worse with every subsequent release. I don’t know what’s going on at Sun but I have an idea: Ian probably did an initial coding for a Debian release, got yelled at by Jonathan, and ended up using Sun code, which is so slow to boot up (Live! or Solaris) that’s it’s not even functional on modern laptops. Forget Sun and move on with BSD!
OpenSolaris has always been based on Solaris code for networking and more. If you’re having networking issues with newer releases, please file a bug.
My personal experience has been quite the opposite. Networking used to be spotty starting with the 2008.05 releases, but by 2008.11, all of my networking issues had been resolved.
I love Nexenta as a server OS (when tools like ZFS and zones come into their own), but never really saw the advantage of running a Solaris kernel on a desktop when there are other, more mature, desktop distros around)
I understand that OpenSolaris is Suns attempt to bring new developers into the fold, but people need to remember that Linux took years before it was practical as a desktop OS
True, I think it was the Year of the Linux Desktop (2003) that it really started to hold its own.
So what was the reason for creating OpenSolaris? Noone ever said that it should be a general purpose desktop competing with Windows, like say Ubuntu, but rather a system for developers and admins, who are dealing with Solaris infrastructure and students who are looking for an easy way to get started.
Instead of comparing desktop experience with Linux (I’ve used OpenSolaris for some time as my main OS and it didn’t feel so bad at all.) you should rather check out some great technologies that are included, like the awesome new network virtualization, Solaris process scheduling and resource management or the storage framework comstar, which allows you among other things to turn your server into a fibre channel array. You don’t have any of these things in Linux. (While we’re at it, Linux doesn’t even have a real fibre channel stack and Linux’s resource management is very incapable.)
I would suggest writing articles about stuff you are competent with or at least taking some time to research.
As we all know this is only the second attempt of the so called Project Indiana. It is in fact still a tech demo.
I think in the future we will see a much better and user friendlier version coming out.
For the time beeing stick to your favorite Linux Distro, i.e. Ubuntu, Debian, ZevenOS. They are much faster and offer more for a normal desktop user.
OpenSolaris is a nice to have tech demo and testing platform.
The reason it sucks as a desktop distribution is that it isn’t a desktop distribution. I’d love to see OpenSolaris as a server distribution, I don’t understand why Sun managed to make such a mess out of a project that could bring a really nice server experience.
Like, say, Fedora Linux, it is really meant to be both, and I use it as such. I can watch videos, listen to music, use all the software I need to use except for a few games (if I had more time these days I might even be able to futz around with Wine enough to get them working), and I have ZFS which I benefit from greatly. I haven’t really used dtrace yet (not enough time), though I’ve wanted to a few times when I’ve hit a debugging snag.
I also liked SXCE more than the initial release of OpenSolaris, but with 2009.06 I’m starting to come around. I wiped my SXCE install (keeping my raidz pool) and replaced it with OpenSolaris; I was back up and running in 2 hours.
As for all of the people hating on the desktop experience, I think it is as good as a Gnome UI could be. Sure, linux works with more hardware, but if you have the right hardware (which in my experience isn’t too hard to come by, including laptops), then it is great.
OpenSolaris is a kernel plus apps much like openSUSE or Fedora. At the heart of OpenSolaris is the SunOS base which is essentially the kernel, much like Linux sits at the heart of openSUSE and Fedora.
So… OpenSolaris is NOT a Linux distribution because it does not use Linux.
Remember, that some of your favorite Gnome and KDE apps work on Windows… that doesn’t make it Microsoft Linux… right?
SunOS/Solaris have been around for quite some time. Solaris I think first came out in 1988, 3 years before Linus had something. And before than the kernel that Sun used was BSDish… and dating WAY back.
So… is OpenSolaris a Linux… NO. It would be better to consider it another *ix out there in the world.
The reason why HPUX and AIX can run circles around Solaris/OpenSolaris is because they are both SVR2 derivatives, whereas Solaris was a SVR4 derivative (back when Sun and AT&T were trying to take control of all of Unix… leading to the formation of the OSF, etc, etc.).
One could argue that OpenSolaris has more contemporary kernel features than their HPUX, AIX, etc. bretheren, with perhaps the exception of SCO Unixware (but that thing has a LOT of warts… hard to see anything worthwhile).
OpenSolaris is just another OS… like BSD, like Windows, like AS400, etc. It is Unix, so it compares well to other commercial Unix vendors and since Linux is very Unix-like, it compares well with it and even with the shell and GNU applications (and even more so since OpenSolaris has replaced a lot of the Unix standard apps with their GNU equivalents).
Personally, I wouldn’t choose OpenSolaris just because they sort of gave up on being good over the past few years. Thus they are now part of Oracle…
With that said, I do have access to OpenSolaris where I work… it no longer has a root user… which is different, it has a strange init (actually present from 10+)… but all in all, it’s not horrible.
I doubt I’d consider it for a desktop though. The Sun approach has always been to PAY for their drivers… the whole GNU/freedom thing is still pretty foreign to them. And therefore, a Linux based distribution will likely work with tons more hardware out the box than OpenSolaris.
I use OpenSolaris here and there. And it works for me. And because I am not alone who thinks this, OpenSolaris has a reason to exist.
I couldn’t care less what Thom thinks of it. It serves my needs.
Most Linux and Solaris systems are used as servers. I prefer a server that is good at doing server things, to server having a good desktop, and OpenSolaris is a very good server.
Any operating system with a monolithic UNIX-based architecture is so superb for servers.
I use Solaris and OpenSolaris is like a slug nailed to the table. I much prefer Solaris. It’s boot times aren’t fast but runs fast once booted.
I have a ZFS based home server for an year or so that acts like samba/afp/torrents/firefly/virtualbox/ipcam security system. without even single failure. The place I live has a frequent power failures. I’ve had different kinds of systems before. that were corrupted on each power failure. but with latest 2009.06 OpenSolaris I’ve had no problems so far. before I had Solaris 10 based on UFS which worked ok but then one day. filesystem was corrupted. So I gave OpenSolaris a shot. it was version 2008.05 as I remember. IPS problems and things like that. but overall was ok. 2008.10 delivered all the bugfixes I need for IPS to work correctly. and 2009.06 is rocksolid release. it’s almost amazing. yes configuring such system was hard enough considering that I had to build much software from sources. but again when it’s configured properly it just rocks. Some of you might say that I need UPS system to withstand power failures. but is there any UPS system that can backup home server for 30 minutes and sometimes even more? This system is power failure proven and is the best choice for a home server
[q} Some of you might say that I need UPS system to withstand power failures. but is there any UPS system that can backup home server for 30 minutes and sometimes even more? This system is power failure proven and is the best choice for a home server [/q]
You usually don’t need 30 minutes of UPS backup, you need enough time to shut down your system safely in most cases 5-10 minutes or so is enough. Most UPSes will shut down your system automatically if you have a power failure longer than 5 minutes or so.
Even so, ZFS is nice especially if you have large disks, as you don’t need to fsck at regular intervals. a procedure that could take several hours on a few TB of ext3 formatted disk
I recently gave OpenSolaris a go and I have to say that I had the opposite opinion to you.
The good:
* My install was smooth (yet a little slow) but it was fairly easy to get it running on my machine.
* In general I found the OS to work extremely quick on my Core 2 Duo, 3GB RAM, 512MB GeForce 7600 GO notebook, it was simple to enable 3D effects like Compiz, etc, the NVIDIA drivers were built in.
* In general the OS was visually pleasing to use, I like the default theme, other themes like clearlooks, lots of wallpapers, nice icon set and I like the default set of Compiz effects.
The bad:
* It really did a number on my Windows installation, there was no option to install the boot loader to the partition that OS was installed on to. I had to use recovery console to get my Windows 7 installation back again, and instead of it being straight forward like Linux where you just choose “Startup repair” I had to use diskpart to re-enable partitions, as well as rebuild the BCD info, etc, it was painful (I did get Windows back but I know LOTS of people will give up in frustration and probably reinstall).
* Wireless liked it inexplicably drop out
* Package manager isn’t very nice and it really slow
* Lack of software but I don’t blame Sun for that
Overall I wish the project the best of success and I am sure things can only improve. Initially I am impressed but since it’s young I won’t make any personal final judgment on it.
I wasn’t impressed at all. Maybe it’s because I expected so much given Sun’s history, but I was very disappointed.
I will say that Gnome seemed somehow more uniform and cohesive than on any regular Linux distribution I’ve ever tried. That was the one and only good part of my experience though.
I ran it on a 2.2GHz system with 1GB of memory and a fast SATA drive. The video was onboard Nvidia 8100 on the board’s PCI-E bus. You’d think this would be a great test system but it was slow on installation and boot relative to Slackware and Xubuntu (my two favorite distros). Once up and running, speed wasn’t really an issue, and X and Gnome were very responsive. It felt like I was on a BSD system. The preinstalled proprietary Nvidia video driver was a nice touch.
However, neither my onboard nForce NIC, nor my Linksys wireless PCI card were detected. I scrounged up a 14-step process to enable ethernet using a custom driver and was able to get the OS to see the NIC, but I never was able to get an IP address or internet connection. Sound was hairy too; the mixer either failed to load, or loaded and showed no active devices. I could detect a faint sound but couldn’t change the volume at all in the OS. I had to resort to using my speaker’s volume knob and ended up with overdriven static.
It was at this point that I felt like I’d stepped into the past, to my Linux beginnings and Red Hat 6. Nearly nonexistent out of box driver support, and while the GUI was great, it was a nearly unusable system without network access. UNIX and its derivatives are network-centric operating systems, after all.
I had high hopes and I suppose that’s a major part of my disappointment, but nothing can change the fact that it was a poor user experience for me.
Firstly, …
My computer system is a HP XW9300, having dual-slot single core Opteron 2.4 GHz CPUs, 4GB RAM, 1st generation Western Digital Raptor HDDs.
Two operating systems are installed being
– 64-bit OpenSolaris (patched using “dev” repository)
—> “uname -a” gives
SunOS Argo 5.11 snv_117 i86pc i386 i86pc Solaris
– 32-bit Win2K
and these operating systems are accessed using the simple GAG boot manager.
Secondly,
OpenSolaris as well as Win2K are fast on my system.
The 64-bitness of OpenSolaris truly pays off well. e.g. OpenOffice and Gimp on OpenSolaris show marked performance improvement over the Win2k versions.
Thirdly, …
My OpenSolaris boot time (i.e. from GRUB screen to Gnome login shell) is about 35 to 40 seconds. Shutdown is about 8 to 10 seconds. For me, this is fine for an operating system have many high technology “internals”. Any “serious” issues I have found with OpenSolaris (more so with the “dev” repository) has been boot-related and for some systems you may need to investigate the GRUB options. Remember, all hardware is not the same. For me, I found the “avoid PCI reprogramming” GRUB option to be useful. I avoid the text splash screen during boot and just have the conventional white text on black background so that I can see boot related messages to facilitate my timing of boot period. Also, I have about 10 boot environments (since I evaluate long term stability of patches) and the system is really fine.
Fourthly ….
My OpenSolaris box is a C++ development box, using SunStudio tools. I patch using the “dev” repository and the system is stable, quick, and fun to use. My primary criteria are performance, stability, and standard “ease of use” and redundant/excessive/* “eye candy” has no priority.
Fifthly, …
My backup OpenSolaris box is a dual core Athlon 2GB RAM system and this box is fine. I recently installed 32-bit OpenSolaris on a Pentium IV (3 GHz) system and found boot and application loading times to be much slower than the AMD systems I have. Then again, AMD has the direct connect architecture for several years in which Intel recently (2008) had cloned this idea and implemented in their CPUs.
Sixthly, …
People should not forget the origins of (Open)Solaris and Linux. While Linux was growing from someone’s bedroom and initially penetrating the commercial scene through the back door (before Linux’s “saviour” IBM apparently woke up and started supporting Linux, rather than supporting it’s own OS/2), Solaris was satisfying real world constraints. The Solaris kernel has been big-iron (i.e. multi-processor, multi-core) ready for years since the Solaris kernel was designed on that foundation in order to address real world challenges. Linux’s SMP offering has been an afterthought and definitley not as mature.
e.g. Back in the early-to-mid 1990’s my IT admin friend was managing a heterogeneous computing environment for a brokerage house in Sydney (Australia) consisting of Microsoft Windows clients, Novell file servers and Sun SPARC hardware. This SPARC hardware was used to handle real-time shares/futures processing. During the design phase of this real-time framework the company had prepared a specification for the target performance/stability of the proposed framework (type of operating system was less of an issue, as long as it could be interfaced to the rest of the network). It was Sun with their multiprocessor SPARC architecture, not IBM, not HP, etc,. who were only able to give the required quality of service guarantees.
The OpenSolaris codebase is mature w.r.t. x86 and SPARC CPUs; x86 support since early 1990’s.
Rather than people being petty with subtle dramas, they should realise that a good thing Linux did was make Sun go back to it’s roots and release a capable operating system as Solaris as opensource. I have been a C++ developer for Linux/Windows/OpenSolaris platforms and when gauging these platforms in terms of C++ tools, console/terminal functionality, and operating system features then OpenSolaris has been the big winner for me. I have introduced OpenSolaris to friends who I helped switch from Windows to Linux and they have been impressed with OpenSolaris (they have modern AMD boxes).
Some people say words to the effect …
“so what if Solaris has great technologies like ZFS, DTrace, good multithreaded/multiprocessor ability, predictive self healing, package management with rollbacks, etc., but that is more good for the server and not that important for the desktop …”
to try an indicate that OpenSolaris is less suited for the desktop.
What people should realise is that these “server-like” features are able to run admirably in desktop/workstation (Open)Solaris systems and has enormous benefit to developers like myself. The benefit is that nice operating system features exist to complement the user/developer/admin work. I now have an arsenal of great technologies/tools which allow me to easily approach the goal of developing good quality software code since the features of the technologies/tools allow me to interact with the software development/testing process in ways which were not possible in previous years (Linux/Windows); e.g. DTrace embedded in the programming tools so I can run “experiments” on my compiled/linked executables.
Any serious Solaris-based developer realises SunStudio, not gcc/etc., is the better way. The enhanced development environment offerred by OpenSolaris is a bonus for the user of OpenSolaris applications.
(Open)Solaris is a UNIX as it complies with the OpenGroup’s UNIX accreditation. Linux is a UNIX clone, as it’s “too-dynamic” development model does not allow it to satisfy a UNIX accreditation; it’s too much of a moving target and unlike (Open)Solaris, less priority for maintaining backwards binary compatibility (ABI stuff). This is important because UNIX systems reflect a maturity in design of the underlying API’s and subsystems. It’s easy for Linux to run away and hap-hazardly support many themes since it has no mature/accredited framework that has to be maintained to ensure an accreditation is always applicable.
People might knock OpenSolaris for lagging Linux’ hardware support but then again if hardware support was the dominant issue then we’d all be using Microsoft’s products. At least the OpenSolaris harware support is getting better.
Also, it’s very convenient for certain Linux supporters to forget that Linux (as well as other operating systems) have attempted to clone Solaris-related technologies (e.g. ZFS, DTrace) with limited success. Sun have a good set of competent Solaris engineers who kick started the opensourcing of Solaris and help maintain the innovation that is characteristic of the OpenSolaris community. OTOH, we have IBM who ditched the OS/2 operating system which had great potential and, like HP, has embraced Linux but still maintain their proprietary UNIX operating systems. Obviously, accredited UNIX systems offer users certain experiences that Linux does not, but the important issue is that Sun opensourced it’s UNIX operating system while IBM and HP did not. I only stress this point since on one hand IBM/HP imply “Linux for the masses” and on the other hand they imply “Linux has issues and so some people need to stick with our closed/proprietary UNIX offering”. At least with Sun, they are implying (Open)Solaris can be used everywhere. If x86 does not scale well for your needs, then SPARC (and even some mainframe tech) is there for your OpenSolaris needs.
Lastly, if you wonder about the potential of (Open)Solaris then take a look at pages like the OpenSolaris community pages at
http://www.opensolaris.org/os/communities/#portal
The OpenSolaris community is coherent and adequately structured to enforce the ongoing evolution of (Open)Solaris.
I believe this aspect of OpenSolaris would scare many of OpenSolaris’ “vocal” opponents. If it hasn’t, then they should be scared.
It seems like you are arguing more for people to respect Sun and their work, rather than why anyone other than a Solaris developer would want to run OpenSolaris.
I’m mostly inclined to agree with all that, but it’s too late, mate. SUNW is no more.
Perhaps more on topic, as a Linux sysadmin and general fiddler of shiny knobs, the desktop distribution is an easier way for me to try and wrap my brain around Solaris and is certainly all around more pleasant than any previous release of Solaris/Intel. As more of the production installed Solaris base moves to the new stuff, it’ll be better for everyone, but in the meantime … it’s been nice to try out zfs and smf and such.
The network driver situation has mostly prevented me from trying to use an OpenSolaris/Nextena/StormOS install for anything, but I will keep trying the new releases as I do want to see how it comes along.
Greetings adricnet.
“It seems like you are arguing more for people to respect Sun and their work, rather than why anyone other than a Solaris developer would want to run OpenSolaris.”
–> Yes, it appears I got abit emotional from the developer point-of-view. More broadly speaking, I feel Sun has the innovation “bug” and it’s good that a charged-up community exists to allow us to have access to an alternative operating system that has interesting (and even industry-setting) features. I suppose what I failed to stress was “watch this space” as I hope/expect (Open)Solaris to be making waves in the future and also that Solaris has a proven commercial track record (mostly server space ?), implying a mature platform that any user should seriously consider as a “bonus”, that may only need to be morphed here-and-there by the community (as time goes on by) to reach an ideal desktop/workstation experience.
Also, “Sun and their work” have stood out for me since I keep remembering years ago as an OS/2 user when IBM let the OS/2 platform rot. I would have liked to see OS/2 supported as a mainstream OS up there with the likes of Solaris,AIX,HP-UX,etc. Can you imagine if the relatively mature OS/2 was released as open-source and even one tenth of the Linux opensource effort was applied to OS/2 … would have been interesting.
Reading info on the enthusiastic OpenSolaris community reminds me of what OS/2 wasn’t but may have been if BIG BLUE had “liked” it’s own product line rather than simply selling out and going for the quick buck with Microsoft product sales. As, such I am suspicious of IBM even with it’s recent “love” for Linux.
“I’m mostly inclined to agree with all that, but it’s too late, mate. SUNW is no more.”
–> Yes, I would have liked Sun to be independent, more “pure” and free from potential contamination. We’ll see soon what’s Oracle’s deal.
“Perhaps more on topic, as a Linux sysadmin and general fiddler of shiny knobs, the desktop distribution is an easier way for me to try and wrap my brain around Solaris and is certainly all around more pleasant than any previous release of Solaris/Intel. As more of the production installed Solaris base moves to the new stuff, it’ll be better for everyone, but in the meantime … it’s been nice to try out zfs and smf and such.
The network driver situation has mostly prevented me from trying to use an OpenSolaris/Nextena/StormOS install for anything, but I will keep trying the new releases as I do want to see how it comes along.”
–> Let’s hope (Open)Solaris get’s more shiny as the future arrives.
ZFS only provide snaphost of file, whereas timemachine is more a backup system ( I think it requires a separate disk ).
ZFS snapshot doesn’t save you from backup ( time machine is some kind of backup )
I would say the 2 are a poor man’s versioning system.
So for a good workflow, have a versioning system ( which is complicated and is bothersome for the mundane user ). And for your data safety have a backup system.
I think timemachine wins on this one for its simplicity and integration into the OS. ( I need a second opinion about windows shadow copy )
Time machine doesn’t require another disk, Apple set up the program to make you believe you need to buy their time capsule. But you can use a regular external harddrive for alot cheaper. However, to set it up you have to go to the command line. Funny, how they got it set up that you have to go to the command line to set up an external or use the same harddrive for time machine. But then again, I am amazed they haven’t gotten hit with an Antitrust case by now.
I have had the same experience when I tried OpenSolaris (OS) 2009.06, i.e. it is very, very slow in every aspect. Additionally, since I’m new to OS world, I needed some documentation, and to my surprise there is very little documentation! This is the biggest problem for new OS users!. I also don’t know why OS takes nearly 800MB of ram, even though I just started it.
The 800 MB of RAM is because of ZFS. ZFS is supposed to be wonderful because it caches lots of stuff in RAM, but unlike normal cache, ZFS cache isn’t freed when applications need more RAM… It just stays there. What you get is an utter RAM hog.
As for the sluggishness… Beats me, maybe it’s just slow because XOrg and Gnome are slow. Last I tried it it didn’t seem much slower than most of the desktop Linux distros, except in terms of boot time.
Pardon, but this isn’t true. Please cite your sources if you have proof otherwise.
There is actually a lot of documentation. Have you ever visited docs.sun.com?
What about the OpenSolaris-specific documentation in 11 languages here?
http://dlc.sun.com/osol/g11n/downloads/docs/current/
Find out more here:
http://www.opensolaris.com/learn/
Positives:
– ZFS – this one is hard to beat. It just feels modern
– VirtualBox – Linux has it too but I come from *BSD (which only recently ported to FreeBSD).
– Zones which is easier than FreeBSD’s jails
Negatives:
– mess between amd64 and i386
– linking problem between gcc and sunstudio (especially when compiling from source)
– feels like Linux (no separation between -core- userland and applications)
??:
– SMF: powerful but complicated.
Six months later (after one successful upgrade, 2008.11 => 2009.06) I decided it’s not worth the effort – uninstalled and replaced with FreeBSD 8.0-BETA1. My sanity is back
(I just remembered – OpenSolaris’ Zones are much more powerful than FreeBSD’s Jails. One outstanding feature is its ability to run various versions of Solaris and CentOS. Yes – you can ‘install’ and ‘run’ CentOS with native speed on OpenSolaris. Minus epoll though)
Edited 2009-07-21 04:01 UTC
Well … all what I have seem is complaining and complaining and complaining … All say linux is better linux this and linux that but it seems that a very small group of people remember how linux was … let me enlighten you a bit.
I started Linux a bit late with redhat 5.2 … X was … well … it was a crap. To configure X we needed to known every attributes our Graphics and monitor had, including mouse and/or keyboard. It was considered “geek”.
for me OpenSolaris nowadays is the same it is to be used by those that know Unix and for those that want to try and learn OS mostly. Making another comparison, I learn most about OS with Gentoo when they had stage 1 install …
Moving on … yes ZFS uses a lot a mem, yes timeslider is not nothing more and nothing else than a zfs send and a zfs receive, and yes it is a bit slow booting.
This is my experience with SXDE which I use since b66 now b117. It had it’s issues but the ability to make a live upgrade while I’m working it’s just awesome (other os’s might have it also, don’t know which), in the end of the upgrade and keep working and in the other day I boot into the new environment.
All my work and normal usage, see movies, listen to music I do with with SXDE. Games I use windows, why? well I’m not rich and I’m not getting a workhorse computer so I can run my games under wine … bah.
I’ve seen OpenSolaris working in HP, Toshiba and Dell without problems … I have my Huawei E220 working with OpenSolaris.
So … as to my experience and with 3G’s RAM I can have Firefox 3.5 with over 15 tabs opened … thunderbird … staroffice 9 … VirtualBox with two VM’s working W7 and Fedora 11 each with 512Mb without it using swap at all. The difference for 2G is the Vm’s nothing more.
If I want to play with iscsi I do it, if I want to use crossbow, easy just use it.
By the way … Slowlaris in networking … well … that’s outdated … if you are a european it’s like saying Alfa Romeo stills has electric defects … people tend to maintain some phrases without even prove if it’s still real.
X works like a charm with compiz also, even with a quadro fx from the laptop it’s fast.
Anyway … just spewing out the barbarities that I think are.
Maybe I’ll get my ass wiped
Tom,
I tried it when it was first released. In my working days, Solaris was an OS that was held in the same esteem as IBM’s MVS and VM (much better than DEC’s VAX/VMS (I think it was called; the forerunner to Microsoft’s Windows NT :-)) – that will get some negative comment). So I was keen to try it on my PC desktop. I agree with your findings, Solaris took hours to install but only minutes to replace (with Ubuntu).
Can I suggest that you try PC-BSD, it seems to be a very robust desktop OS and has, IMHO, the best package manager there is.
Regards,
Peter
Edited 2009-07-21 05:51 UTC
We’re comparing OpenSolaris here, not Solaris. OpenSolaris should install in 15 minutes or less. So your comment and experience don’t quite apply.
“We’re comparing OpenSolaris here, not Solaris. OpenSolaris should install in 15 minutes or less. So your comment and experience don’t quite apply.”
BinaryCrusader (what an odd pseudonym), you ought to read the posts before you dive off into commenting. There is a short initial sentence followed by a second sentence that shows (or tried to show) my original good feelings towards Solaris and hence Open Solaris. These original good feelings were then changed by the Open Solaris experience.
I’m sorry if that is too difficult to understand.
The world would be a nicer place if we occasionally thought first and then acted; don’t you agree?
Regards,
Peter
>The Linux desktop [..] It’s a fully usable, stable, and secure operating system that can be used quite easily by the masses
I disagree. Linux is getting more and more unstable. In the last few years, it has grown in size, weight and unstability. It’s looking nice every day – sure – but it’s also behaving more and more like Windows.
Which is one of the few reasons I hear more and more people trying out other OS, such as Solaris. On that part, Solaris’s stability is unquestionable.
But, true, Solaris is not for the masses.
>it would be “slow”.
Well, if you use Gnome & Compiz, I’m not that surprised. But I’m sure you’d be surprised to see what a desktop can look like with FVWM for instance. It can be as nice… but much lighter.
Also, have you checked hardware warnings ? Because I believe that if you find it slow, something’s wrong with drivers, although for instance, nVidia cards are usually very well supported.
Anyway, your post is putting aside Solaris/OpenSolaris’s advantages to quickly.
– ZFS, and the time slider (snapshots) are really awesome ! I just *love* it and all my Linux colleagues are envious, they just tried OpenSolaris.
– stability: it does not crash, does not slow down, and there’s plenty of system admin things you can do without rebooting. Linux is getting into that Windows habit of asking you (or having you…) to reboot more and more.
– documentation: this is just so great, when something does not work, man xxx and it explains it all. Yeah, I’m sure you do that on Linux too, but on Linux, the explanation is wrong or does not work 1 times in 3. In Solaris, it behaves the way it is explained. Things are deterministic. I like that.
– at home, two of us are working on the same Solaris host (a less powerful one than yours) and it speeds…
– it powers off in a few seconds, I just can’t believe how fast it goes.
So, well, perhaps OpenSolaris isn’t for everyone, but I can guarantee it’s pretty cool and wish it good luck !
Your post raises the old question in my mind that is usually raised by many of the other works – are the OpenSolaris folks trying to reinvent the wheel?
/Rant mode on/
This is nothing personal but if there’s one thing that gets my goat it’s this anodyne trotting out of ‘are you/they trying to reinvent the wheel?’; it’s one of the most destructive of rhetorical devices and i wish it could be banned;
a) if someone had not spent a lot of time reinventing the wheel, we would not have spokes or pneumatic tyres, and we certainly would not have F1 racing…
b) The Hovercraft could be said to be a ‘reinvention’ of the wheel, in the same way that helicopters reinvented the fixed winged plane.
Let’s look at basic functions and purposes; we may all in the end be getting from A to B, and some of us may want to fly as a means of doing so but does that mean we should sit there, looking at the birds, because there’s no point in reinventing what your average winged creature does. Did nature leave the bat wingless because there were already feathers?
If you mean, “I have this and it suits me but hey, you are free to rustle up something different that suits you better”, then fine, but usually the import is, “Hey, I reckon what I have will do everything so why are you bothering”. The comment is superficial and superfluous at best, and is singularly negative at worst.
/Rant mode off/
“Let’s look at basic functions and purposes; we may all in the end be getting from A to B, and some of us may want to fly as a means of doing so but does that mean we should sit there, looking at the birds, because there’s no point in reinventing what your average winged creature does. Did nature leave the bat wingless because there were already feathers? ”
Claps hands for you! You have made an excellent point about innovation and creating new ways. Thanks for the post!
For me, OpenSolaris 2009.06 installed in 15 minutes. It boots in 30 secs and poweroffs in 5-10 seconds. It is well known that OpenSolaris has limited hardware support and it will not even install on some configuations. I suspect the reviewer has some hardware that OpenSolaris dont like and therefore behaves strangely. For me OpenSolaris is fast and snappy. I suspect that if the reviewer tried OpenSolaris on other hardware, he will find out that it behaves just like an ordinary Linux. I promise you that if OpenSolari sucked on my hardware, I wouldnt use it either. Im not a masochist. I would have prefered Ubuntu then.
I have now automatically upgraded OpenSolaris 2009.06 to build 117 via the “Update Manager” and everything is smooth and fast. One of the things with ZFS I love is the BE (Boot Environments). Before upgrading or installing a package via IPS, ZFS automatically takes a snapshot, i.e. BE. All these BE will show up in GRUB. If you find out that your new update/installation breaks something, you just choose to boot from your old BE in GRUB. You can have many BE to choose from. One with Oracle vXXX and another with vYYY etc.
When ZFS snapshots, it just writes all data to another place on the hard drive. All old data are intact and untouched. Now you choose via GRUB which BE to boot from. This makes it safe to try out anything, and to rollback via a reboot, in GRUB. This is a killer feature. Imagine a server and you have to do a rollback. Just reboot and everything is identical to as it were.
And for ZFS, it is a memory hog. It grabs all RAM it can get, as a cache – but immediately releases all RAM when asked for. On a smaller system, there are surprisingly little disc activity with 8GB RAM. ZFS is a Enterprise Filesystem and it wouldnt work if ZFS refused to release all RAM when asked for. Stating otherwise if FUD. But it is nice to see that SEGEDUNUM nowadays have stopped stating that ZFS requires several GB RAM. I and others here, have told him many times that 512MB RAM suffices, but to no avail. But now at last he seems to have understood. Better late than never.
Anyway, the MAIN REASON to use ZFS is one and only. It protects against silent corruption. ECC RAM corrects flipped bits caused by radiation, power spikes, bugs, etc. ZFS does exactly the same thing, it protects the data against flipped bits. No other filesystem/solution does that. fsck checks only the metadata, the data is never checked. Hardware raid doesnt fix this problem. For instance, HW raid would never ever never ever fix this problem, described in this short text:
http://blogs.sun.com/elowe/entry/zfs_saves_the_day_ta
In fact, a modern hard disk dedicates 20% of it’s surface to error correcting codes, and still there are errors that it can not repair or even detect. The chance of getting an errorneous bit is constant, albeit very low approximately 1 bit will be wrong in 3 x 10^7 bits, according to physics centre CERN:
http://storagemojo.com/2007/09/19/cerns-data-corruption-research/
The larger the drives, the more bits you will read and soon you will have read several batches of 3 x 10^7 bits. The more bits you read, the more often you will get silent corruption, which the hardware doesnt notice and doesnt tell you. In today’s large RAID, you will always get flipped bits and it will only get worse as discs increase in size. Therefore raid-5 is soon obsolete (there are too many bits involved):
http://blogs.zdnet.com/storage/?p=162
However, ZFS fixes all these problems. It is designed to fix this kind of errors, among others. ZFS chief architecht:
http://blogs.sun.com/bonwick/entry/zfs_end_to_end_data
“The job of any filesystem boils down to this: when asked to read a block, it should return the same data that was previously written to that block. If it can’t do that — because the disk is offline or the data has been damaged or tampered with — it should detect this and return an error.
Incredibly, most filesystems fail this test. They depend on the underlying hardware to detect and report errors. If a disk simply returns bad data, the average filesystem won’t even detect it.”
This is THE text on ZFS to read, by the ZFS chief architect:
http://queue.acm.org/detail.cfm?id=1317400
“one of the design principles we set for ZFS was: never, ever trust the underlying hardware.”
We also have to remember that OpenSolaris is very young and it evolves very quickly. In short it will be on par with Linux, I speculate. Should you have tried a totally new Linux distro that is 2 years old, maybe you would have problem too?
But when all works, and the hardware is supported, OpenSolaris is a very pleasant experience.
Been playing with OS ever since its first release on a developer notebook, though I overally like the platform, so far I yet have to find a reason to really use it in day to day life. Outline: I’m a Java developer, heavily using IDEs like NetBeans or Eclipse and app servers locally installed for testing purposes. Observations:
– Generally, in virtually everything I do OpenSolaris is considerably slower than a trimmed-down, optimized Xubuntu installment on the same hardware. I have yet to find why it is this way.
– Package management: IPS is a pretty good thing IMHO, but in terms of performance it is somewhat behind dpkg or rpm, and, worse: The “official” IPS repositories are somewhat limited compared to, say, Ubuntu universe, and adding blastwave packages seems to pull half the system (GNOME libaries?!) once more to some arcane location (/opt/csw?). Package naming is completely obscure, in most situations (SUNWwhatever – just try to find “the gimp” in IPS repo…).
– These issues aside, OpenSolaris is not really worse than Xubuntu, but, and this is what is important to me, it’s also not “better”, it doesn’t unfortunately provide no real benefit in day-to-day use to consider using it in everyday work. Sure, it has an astounding set of features worth looking at (ZFS, DTrace, Zones, SMF, …), but most of them (at least to me) seem completely meaningless on a desktop system. Eventually, Time Slider is the only “Solaris thing” on an otherwise pretty plain GNOME desktop, and exactly this is a feature so far I never needed (and if I actually did, I could go with TimeVault in Ubuntu having the same effect).
My bottom line is: OpenSolaris, as a desktop, needs to have more “desktop things specific to OpenSolaris” to be interesting. On the server, it surely does excel. On the desktop, there’s IMHO no real need for it right now…
Really? Have you read what I wrote about BE, two posts above yours? And you can install whatever you want, and if it breaks anything you just rollback in an instant via snapshot. This way you can have an environment for Eclipse java 1.5 in one BE in GRUB, another with Eclipse Java 1.6, Netbeans, etc etc etc.
This is not useful for desktops, you think??
Not sure, but at the very least for keeping different Java environments separated, I don’t really need boot environments (although they are a nice feature, no doubt about that). Been working with OpenSolaris 2008.x for about two months in total in production work, trying to figure out whether this is what I want, and never found a real reason making me thing “wow, now _that_ is amazing a thing to have…” – on a notebook, not on the server.
Overally, stating this again, comparing the “experienced negatives” (overall performance, IPS issues, eventually lack of pre-compiled software) with the set of additional features of some use _to me_, I don’t really see a reason to switch to OpenSolaris as my main OS. I’m still cautiously enthusiastic about OpenSolaris, both hoping for it to actually grow a real “community” (rather than just being a “public Sun code repository”), maybe for becoming the first operating system to be GPLv3 licensed (yes, this matters to me), maybe for coming to life in a more “lightweight” incarnation (StormOS with XFCE is pretty good about that actually) for desktop or notebook usage, but so far IMHO it is not yet there. YMMV of course.
K.
Ok, it is your choice, if you do not care about keeping your data safe (ZFS), or making your programming easier (DTrace).
But just one thing, please just skim this articles and see if you really really really do not think that OpenSolaris is totally awesome. Because, this stuff no other OS can do. This is totally unique for a developer. I can not really understand why you would think this stuff not the-coolest-thing-on-earth-right-now!
PHP development
http://blogs.sun.com/bmc/entry/dtrace_and_php_demonstrated
Rails
http://blogs.sun.com/bmc/entry/dtrace_on_rails
Java
http://blogs.sun.com/ahl/date/20050418#dtracing_java
Java Swing
http://blogs.sun.com/bmc/date/20050418#your_java_fell_into_my
And this is possible for other languages as well. DTrace helps Solaris kernel developer tremendously to iron out the bugs. People are in total awe of DTrace (and ZFS, Zones, etc etc). Maybe you non-solaris users fail to see why we are in total awe, or maybe we solaris users are very easily impressed:
“I looked at one customer’s application that was absolutetly dependant of getting the best performance possible. Many people for many years had looked at the app using traditional tools. There was one particular function that was very “hot” – meaning that it was called several million times per second. Of course, everyone knew that being able to inline this function would help, but it was so complex that the compilers would refuse to inline.
Using DTrace, I instrumented every single assembly instruction in the function. What we found is that 5492 times to 1, there was a short circuit code path that was taken. We created a version of the function that had the short circuit case and then called the “real” function for other cases. This was completely inlinable and resulted in a 47 per cent performance gain.
Certainly, one could argue that if you used a debugger or analyzer you may have been able to come to the same conclusion in time. But who would want to sit and step through a function instruction by inctruction 5493 times? With DTrace, this took literally a ten second DTrace invocation, 2 minutes to craft the test case function, and 3 minutes to test. So in slightly over 5 minutes we had a 47 percent increase in performance.
Another case was one in which we were able to observe a high cross call rate as the result of running a particular application. Cross calls are essentially one CPU asking another to do something. They may or may not be an issue, but previously in was next to impossible (okay, really impossible) to determine their effecs with anything other than a debug version of the kernel. Being able to correlate the cross call directly to application was even more complex. If you had a room full of kernel engineers, each would have theories and plausible explanations, but no hard quantifiable data on what to do and what the impact to performance would be.
Enter DTrace…. With an exceedingly simple command line invocation of DTrace, we were able to quickly identify the line of code, the reason for the cross calls, and the impact on performance. The basic issue was that a very small region of a file was being mmap(2)’d, modified, msync(3C)’d, and then munmap(2)’d. This was basically being done to guarantee that the modified regoin was sync’d to disk.
The munmap(2) was the reason for the cross call and the application could get the same semantics by merely opening the file with O_DSYNC. This change was made and performance increased by almost double (not all from the cross calls, but they were the “footprint” that lead us down this path). So we went from an observable anomaly that previously had no means of analysis to a cause and remediation in less that 10 minutes.”
EDIT: See DTrace in action on Linux (on top Solaris with Zones a.k.a BrandZ), where DTrace helps spotting less optimal behavior of “top” command:
http://blogs.sun.com/ahl/entry/dtrace_for_linux
Edited 2009-07-21 19:37 UTC
Hi there;
well… you don’t really have to kind of “evangelize” me regarding OpenSolaris… I spent most of my university life working with old Sun / Solaris workstations and for sure are affectionate towards Solaris and in some ways enthusiastic about the possibilities OpenSolaris does offer. And I have to admit that I am using Sun stuff (NetBeans, Glassfish, not talking about Java of course… :>) wherever possible. Personally, as well, I think many of the features provided by OpenSolaris generally are good, but then again, talking about an open source system, are they really tied to OpenSolaris? ZFS so far also does exist as a (fuse) port for GNU/Linux users. Maybe (not sure, though) DTrace also might be ported to GNU/Linux or other Unixoid systems – I’m not sure.
The only thing I know is, off-hand, that Sun in many respects failed about OpenSolaris. Why on earth that strange “Java Desktop System” (basically a modified GNU/Linux) a couple of years ago? Why does it take so long to make OpenSolaris stable? Why is there no “real” developer community around OpenSolaris so far, comparing to GNU/Linux or the *BSDs? Why, talking about DTrace in example, doesn’t OpenSolaris come with a straightforward, powerful GUI tooling for these features to allow (desktop/developer) users to easily get started with these tools? Why, at the moment, is the set of hardware supported by OpenSolaris (being a company-backed operating system) still felt to be years behind what the Linux kernel provides here? Why, to get back to this example, does a system like Debian cleanly and quickly install packages within a couple of seconds or minutes where OpenSolaris IPS still takes rather long to install obscurely named packages to strange places like /opt/csw/ or /usr/gnu? I think that, given some more love years ago, OpenSolaris by now could be predominant. The way it is, right now it has to compete with GNU/Linux on the operating system, not even talking about Windows or MacOS X (which, as I disturbedly had to realize, seems to be the OS of choice amongst most of my Sun contacts… so much for that).
Asides this, just to add another example: When JavaFX was released, I just was into testing OpenSolaris, and I felt enthusiastic about JavaFX as well, just to figure out that – what? A technology released by Sun, in its initial release not supporting the operating system also released by Sun? That’s simply dumb, from a marketing point of view, in my opinion…
So, overally: I hope the Sun/Oracle merger won’t affect OpenSolaris all too much, or maybe a community will be capable of dropping in keeping OpenSolaris running even without Sun being there backing the project anymore. I still see work to be done, and I won’t hesitate also testing out future releases. Let’s see where it’s heading…
I am not really trying to evangelize you. I am merely asking a question; have you ever seen the DTrace stuff being done earlier on any OS? No you havent. And wouldnt you find it useful, working as a developer?
Therefore I am asking. Because I dont really understand when you say that “OpenSolaris has nothing that Linux hasnt and OpenSolaris needs to distinguish it from Linux”. I really dont understand your line of thought? ZFS is unique. DTrace is unique, just as you have read it can do unique things no other common OS has ever been able to do before in history. No OS. And you call DTrace “no distinguishing feature”? You are joking. You have never seen anything like DTrace before in your entire life. Never.
And for instance, when you say that SUN should target JavaFX for OpenSolaris first and SUN is being “dumb” not to do so. Why in earth should SUN target JavaFX for OpenSolaris? The majority of the Java developers work on Windows. In MY point of view, SUN would be dumb if they didnt target the greatest Java market: Windows. First, pick low hanging fruit, and then at last, release JavaFX for smaller OSes. That is sound business strategy and not dumb?
You know, I really dont understand how you think. But that is ok. If you believe ZFS and DTrace are not totally unique and revolutinizing, then it is good for you.
Hi there…
Talking about DTrace, I find this quite useful actually for doing server-sided diagnostics. On the desktop, doing Java development, so far I simply haven’t encountered situations in which using the debugging facilities provided by Eclipse or, even better, by NetBeans did leave me wanting/needing more. So I agree DTrace is cool, but so far I don’t need it on my desktop. If this is subject to change, my operating system requirements will, too.
ZFS indeed is unique and maybe _the_ killer feature I see in OpenSolaris (after all, that’s why people usually are to point this out as the first argument “pro OpenSolaris”…
). Yes, I want ZFS on a file server providing hundreds of gigabytes, or even terabytes, of disk space to a bunch of local users, and doing so without requiring me to worry about things like “how to share this mess?”, “how to do RAID et al?” or “what to do if one drive fails?”. However I still don’t see the benefit ZFS provides on a mobile device with a single S-ATA disk unlikely to be really expanded as 99% of the features ZFS offers simply are lost / not required here. On a notebook which usually is short on resources no matter how new it is, I don’t want to waste resources on features I don’t need. 
Please, feel free to completely read my posts.
I know these features are unique. But they aren’t from a desktop user point of view who just needs some UI to start NetBeans and maybe a web browser and a mail client. From that point of view, OpenSolaris is “just another GNOME based Unix distribution” (and, given I decide to use XFCE which I prefer for various reasons, I don’t even have nautilus and Time Slider anymore, so it’s even more vanilla).
This, overally, gives a Java developer one less reason to even look into OpenSolaris.
Yes it does. Because it is narrow-minded and blind. Let me give you an example: I was doing quite some effort trying to convince my fellow developers that the Sun tool chain (including OpenSolaris) is good if you’re a Java developer. Some eventually installed OpenSolaris to their workstations and also liked what they saw (indeed, running on a workstation which is not a notebook, OpenSolaris is a pretty nice citizen once all hardware is supported). Then, JavaFX finally appears, with the Java developers wanting to have a look. And now, all of a sudden, I am being asked why on earth JavaFX (Sun) atop Java(Sun) is released for virtually anything except for OpenSolaris(Sun)? To all those who, following my enthusiasm and inspiration, decided to use OpenSolaris, this decision now has ended up leaving them incapable of playing with the latest and greatest in Java RIA development just because of this decision. This is dumb, dumb, dumb! If trying to market JavaFX as a “developer tool” _and_ OpenSolaris as a “developer operating system”, OpenSolaris just _has_ to be supported from the very first moment. Of course, one can focus on doing marketing for JavaFX, completely ignoring all the other products the same company is doing marketing for at the same time. But that doesn’t sound very reasonable. And, overally, I wonder whether this kind of (IMHO) short-sighted marketing strategy might have to do with Suns recent, say, “business difficulties”…
Because you just picked some of my statements and commented them without bothering reading all the text I wrote. This is good for you, of course, but of course this way you aren’t likely to understand.
So, again, a simple question. Take some up-to-date GNU/Linux distribution and a current OpenSolaris installment, and just compare the desktop UI and the applications bundled with them (i.o.w. no command-line tools like dtrace and no system infrastructure like ZFS). What, exclusively talking about the _desktop_ sphere, does OpenSolaris offer that the GNU/Linux distribution doesn’t?
“So, again, a simple question. Take some up-to-date GNU/Linux distribution and a current OpenSolaris installment, and just compare the desktop UI and the applications bundled with them (i.o.w. no command-line tools like dtrace and no system infrastructure like ZFS). What, exclusively talking about the _desktop_ sphere, does OpenSolaris offer that the GNU/Linux distribution doesn’t?”
If you compare the Desktop UI and the applications in OpenSolaris to Linux, there is no difference. But that is THE reason why OpenSolaris was being made. It should look similar to Linux, but offer an Server Enterprise OS with unique features for servers and developers.
If OpenSolaris didnt look similar to Linux, then there would be many more complaints. Trust me, this is a fact. For instance, Solaris has a different userland than GNU/Linux and there are lots of complaints “Solaris behaves strangely, it is not Linux”. The answer is: “Correct observation, because it is Unix. Not Linux. Talk to real Unix gurus and _they_ think Linux behaves strangely”. Somehow people think that Linux is the “original” and Unix is an offspring when it is in fact the opposite. SUN has to adopt to the changing market by releasing OpenSolaris. And now when OpenSolaris is similar to Linux, there are other complaints “why does OpenSolaris look like Linux??? I do not want that!!!”. But, yes you do. You want it to look like Linux. Trust me.
As for JavaFX, I dont agree with you. To me it is natural that a company wants to satisfy the majority of it’s customers. For instance, consider these scenarios where we have two different bugs that need to be adressed, one at a time. One bug affects a minority of your customer and the other affects the majority.
1) Address the minority of your customers first, and then take care of the bug that affects the majority.
2) Adress the majority first, and then take care of the bug that affects the minority.
If I were a manager, I would choose scenario 2) first. To me it is obvious, but I understand that there are people not sharing my view. JavaFX arrives first to Windows, then OpenSolaris. This is a correct strategic decision by SUN – and not dumb. In my point of view.
I rather prioritize a majority of people sick in the swine flue, and afterwards tend to the minority which has asthma, allergy, etc – than vice versa. The majority first, then the minority.
Yes. That’s what I mean, and that’s what I think OpenSolaris is – an operating system with a set of incredibly smart and helpful feature for _servers_. Nice to see we agree here.
No. A “real” Unix guru will know how to make her/himself home on be that Solaris or HP-UX or AIX or GNU/Linux or any *BSD system rather quickly. This comparison sort of reminds me of an MSCE used to Windows complaining that Linux behaves strangely because it behaves different to Windows. Whatever.
No. Those who are deep enough into technical details to really care know rather well what Linux is, compared to the Unix history (== an “open source re-implementation” of Unix created in days in which no one eventually dared to think about something like an “open” Solaris). And those who don’t know, the “end users” and desktop users, don’t care anyhow, they also won’t care about MacOS X partly being a *BSD system. They simply want work to be done.
This works out perfectly well as long as you have only one product to sell and not at all worry about whether marketing one product might drive potential customers away from other products you offer at the same time. I don’t even think this would have required to put priority to the OpenSolaris version of JavaFX. But, the message that comes across here is: “You chose our developer operating system? Well sorry to say so but you’ve become a second class citizen regarding our JavaFX technology, but thanks for choosing our operating system nevertheless.” From an operating system customer (and yes, being into OpenSolaris I am _also_ directly a Sun customer), this is surely to influence future decisions in that direction, as I might wonder whether I always will stay a second class citizen when choosing that very system…
So, I think there is a version of javafx sdk for Solaris now: http://javafx.com/
I could also run the demos fine in Firefox on Solaris.
Just a note that I had the same experience, coming from using linux for 10 years now, opensolaris & solaris are very underwhelming.
Its depressing Sun put so much effort and (good will?), into opensourcing solaris only for it to fall pretty flat.
Maybe we can eventually get dtrace & ZFS… okay okay, there are OSS equivalents to those shaping up pretty good too.
Why do you think it fell flat? There’s now 3 releases, with incremental improvement each time. I think you’re also forgetting a couple of things
– Solaris was a significant amount of code to open source – 30,000 individual files, 10 million lines of code. While development did have a ‘community culture’ internally, it was a pretty big effort to move that process externally.
– OpenSolaris contains a huge amount of change to Solaris 10, with a lot of new technologies that are starting to change the way software is delivered to the hands of the user. Yep, IPS can be a little underwhelming at times (purely by the stuff that hasn’t yet been implemented), but its a drastic improvement to what was there before with SVr4 packaging and manageability around software updates.
Why it fell flat?
– Because Thom says so and almost everyone agrees with him!
Discounting hardware and driver issues which I can forgive a non mainstream OS for. The overall experience is pretty bad.
For instance on the live CD it comes up with some gnoem panel error and asks me if I wanted to delete the applet.
The package manager was so slow, it could’t not download GCC, and kept failing, my connection is fast enough so not sure why this is.
For some reason it also had trouble with samba not connecting to a PC that all my other linux boxes could connect to through samba.
Your count of most people, at least as defined as people commenting in this thread, is incorrect.
Out of the people that commented one way or the other on OpenSolaris as a desktop OS, 20 seem to think it is roughly as good as or better than Linux, while 14 people seem to think it is not as good or much worse.
Furthermore, there are a couple of problems with the categories below. Are we talking about people that consider a desktop OS one that is good for desktop use for “average joe”, or one that is good for developers. To me, a good desktop OS for developers should suit both their work and pleasure needs, but it may be more difficult to perform certain operatins than in the default (windows or OS X).
Also, a lot of of folks seem to be trying out Solaris on their old Pentium 3s or 4s, which granted linux performs better on, but if you use a modern system, I think you’ll find there is no noticeable (not to say measurable) difference. Even I don’t use OpenSolaris on an old 512MB laptop I have – I don’t even use Linux. Windows XP (a trimmed down version) is the only reasonable “modern” OS to use on it (I have tried them all, and more).
Nonetheless, I classified the below based on whether or not people said, more or less, if it is a good desktop OS (without adjusting their definition):
OpenSolaris as a Desktop OS is not as good as Linux:
Total: 14
Thom
fretinator
broken_symlink
Leszek Lesner
diegocg
fernandotcl
cjcox
kragil
twm_bucket
Morgan
Marcin
Gullible Jones
ideasman42
gjoahnn
OpenSolaris Desktop is the same as Linux or, it is awesome:
Total: 20
Lazarus
Bill Shooter of Bul
oxygene
binarycrusader
KevinM
Jondice
dvzt
phoenix
renhoek
coolvibe
Troydm
cade
edogawaconan
etherealsoul
axellec
Kebabbert
kawazu
glynnfoste
Kalessin
sorpiga
?? or borderline:
Kishe
scribe
Laurence
unoengborg
DistinctiveWeb
OddFox
dvhh
adricnet
Edited 2009-07-21 19:08 UTC
Well I read most of the comments on this topic and wow. I am a average Linux/ Windows user who has installed several versions of Open Solaris and each time I can say it was the same results, a system that ran slow and a lack of drivers and other apps I can easily obtain in windows or Linux, I prefer Linux over Windows. So I can say for the average user who wants something that just works and does not require allot of tweaking and so on will not go for Open Solaris. I really don’t pay much attention to the kernel, as long as it supports my hardware and as long as the ditro makes available proprietary drivers for my Video card that I do not have to compile myself then I am happy.
One major issue I have brought up before but which seems to get little attention is that you need a support contract to obtain security fixes and other updates for OpenSolaris releases.
Otherwise you only have the option to either use the development version with all its instabilities, beta versions etc. or the stable but insecure and essentially unmaintained release version (note that you also currently cannot upgrade single packages from the /dev repository when running /release).
The cheapest support contract for OpenSolaris giving you access to the support repository is $324/yr. (it includes 48h e-mail response support which is something developers/home users/enthusiasts most likely won’t need).
So I would say that is a pretty major difference from popular Linux distributions such as Debian stable, Ubuntu, Fedora, OpenSUSE etc. which give you free access to securtity and stability fixes of their releases.
That is simply not true.
e.g. http://dlc.sun.com/osol/advocacy/opensolaris-slashdot-session-28-05…
Well it is true. Fact is that 2008.11 contains numerous critical security vulnerabilities e.g. in OpenSSL, ipfilter, or Firefox, all of which have been fixed in the /support and /dev branch only.
The 2009.06 release e.g. contains a Firefox 3.1beta with critical vulnerabilities, a fix is available from /support or /dev only.
For more examples of unfixed security vulnerabilities in both the 2008.11 and 2009.06 release branches check http://blogs.sun.com/security/category/alerts .
Also note their change in policy on their website which I have previously mentioned in http://www.osnews.com/thread?366658 .
Not only does it run like whale on land, it doesn’t even give you security updates for releases.
( http://opensolaris.org/jive/thread.jspa?messageID=384939 )
Just pathetic.
Just use Linux! It is really free, you can get free updates for up to 7 years in some cases and BtrFS will wipe the floor with ZFS.
🙂
In my view, this is non optimal behavior from Linux:
http://opsmonkey.blogspot.com/2007/01/linux-memory-overcommit.html
Linux over allocates RAM. And when it needs RAM, Linux starts to kill processes randomly. This is horrendous, yes? Whereas Solaris doesnt over allocate RAM and lets the processes run till theyve finished. Imagine you have a process running weeks with a calculation and suddenly it gets killed?
Maybe this has something to do with the declining quality of the Linux code? Linux kernel hacker Andrew Morton explains:
http://lwn.net/Articles/285088/
“Q: Is it your opinion that the quality of the kernel is in decline? Most developers seem to be pretty sanguine about the overall quality problem. Assuming there’s a difference of opinion here, where do you think it comes from? How can we resolve it?
A: I used to think it was in decline, and I think that I might think that it still is. I see so many regressions which we never fix.”
Or this thread:
http://kerneltrap.org/Linux/Active_Merge_Windows
“the [Linux source] tree breaks every day, and it’s becoming an extremely non-fun environment to work in. We need to slow down the merging, we need to review things more, we need people to test their […] changes!”
I think it has something to do with Linux not having stable ABIs?
http://linuxdevcenter.com/pub/a/linux/2004/09/02/driver_ease.html
“the incompatibility between different stable point versions of the kernel hampers the Driver on Demand concept. You could compile a driver for 2.6.5 and it would probably not work on 2.6.10 if you simply loaded the precompiled binary module; you would need to recompile the driver for each kernel version.”
That is a potential source of unstability. Your driver seems to work fine, with your new kernel. But in fact, on rare occasions your new kernel + old driver combo will just explode but it seldoms happens so you never make the connection. (Do you always recompile all your drivers when you get a new kernel?) The result is that Linux is potentially unstable:
http://lethargy.org/~jesus/archives/77-Choosing-Solaris-10-over-Lin…
And
http://www.enterprisestorageforum.com/sans/features/article.php/374…
Quite extensive troll collection you have there.
“Quite extensive troll collection you have there.”
Thanx! But it isnt hard to find such links. You know, some of those “troll” links are from Linux kernel mail list, where Linus Torvalds discuss in the thread. Do you call Linus a troll? Do you call Linux kernel mail list a troll site? Maybe you should reread those links again, before you call them troll?
If Linus himself wrote that “Linux has no stable ABI” on his blog then that blog would also be troll? Hmmm?
Kebabbert you’re absolutely clueless. In fact it’s downright hilarious to read your posts. Every post you make is littered with ancient articles that are no longer relevant to the current state of Linux and quotes taken out of context.
For instance the OOM issue has been fixed for quite a while now. It is no longer relevant. The quotes you use to try to justify Linux code quality being poor is talking about the dev tree not the released kernel. As for drivers it isn’t relevant for Linux. The drivers are in-kernel and are fixed by the kernel hackers themselves. The author of the filesystem article you link to at the end is just as clueless as you. He claims that Linux filesystems cannot scale and have poor throughput. He even includes XFS in his criticism! XFS is extremely scalable and high throughput. Claiming otherwise is an obvious sign that you know little about it.
Edited 2009-07-24 14:39 UTC
It isn’t pathetic – I think you’ll see the same things with Redhat enterprise linux, etc.
If you really want free security updates, then you could try nexenta – though I don’t have much experience with it.
Really amazing what people can complain about in such an extreme way…
As a *Desktop user*, are you really that worried about needing all of these security updates? The one exception is probably Firefox, and I had no trouble updating it through IPS to the latest version.
Also, from what I’ve seen, btrfs will give you about the same functionality as ZFS – each fs may have slight performance gains in some areas over the other. I’d like to know how it will wipe the floor with it though
The match for enterprise linux would be Solaris 10, not OpenSolaris which is according to Sun aimed at developers, students or early adopters.
Yes I could, and I can keep running a supported Linux distro as well. The subject of this article was however more like comparing OpenSoaris to Linux distributions.
Yes I am. It’s not only about what vulnerabilities already exist. Imagine a flaw in the SunSSH client or a low level library like libjpeg, poppler etc.
Then I wonder how you did that without upgrading the whole system to /dev or buying access to the /support repository.
IMHO OpenSolaris is technically superior to Linux, the lack of a maintained free release make it an unsuitable choice for their target group.
First, thanks for the no-flames and logical response. I see your point about library problems.
That said, is using /dev really so different than just using Fedora Linux (aside from the fact that Fedora probably has way more people working on packages than Solaris does)?
I am using /dev, and I have been since before 2009.06 came out. I haven’t had any problems related to this. Its a little more bleeding edge than OpenSolaris release, but not much – I think the margine of difference is probably small compared to that of OpenSolaris releases and Solaris 10. And with BE’s, its very easy to switch to a previous version should I wish to.
Edited 2009-07-22 17:25 UTC
Download updated firefox here:
http://opensolaris.org/os/community/desktop/communities/mozilla/dev…
Trolls! Do not feed!
I think people who are worried that Solaris has an uncertain future should read this article:
http://www.computerworld.com/s/article/9131829/Oracle_s_Sun_buy_Ell…
Solaris will continue to be improved by Oracle as a server operating system, that is as sure as anything.
The question of how OpenSolaris will continue to improve in terms of desktop usability is slightly more uncertain: it may not have the backing from Oracle that it had from Sun, but I think Solaris engineers and the OSS community in general will continue to work on it.
So more specifically, the question is – will desktop driver implementation continue at a good pace? This is because writing drivers is going to be more difficult than porting over apps to Solaris and making use of ZFS, zones, etc. I’m not sure how much is currently being done by Sun engineers paid to do driver development, and how much is being done by the external community. If we knew that ratio, then we’d be able to have a better worst-case-scenario of the future of desktop Solaris, which still doesn’t seem all that bad to me: even if most of the driver development is being done by Sun engineers now – I doubt Oracle is going to want to fire them … they are a big reason why Oracle bought Sun.
As always, insider clarification would be welcome.
Some forgot to mention that OpenSolaris is production ready and free.
It has also has crash dumps and panic utilities build in the system to help troubleshoot buggy applications. For RedHat you need to install a netdump server to support this feature.
Have you ever tried to mirror root partition on linux or worked with RAIDs (VolumeManagement)?
Solaris FC and Storage Multipathing Software included with Solaris. Network cards bonding is easy to implement.
Yes, for example when YaST bring up the paritioning during installation you create a RAID array and a root partition on it. Just a few mouse clicks, works like a dream.
Solaris does not work like a dream… It can’t detect all disks on a 24 port controller because of stone-age limitations in its SCSI subsystem.
Sure it does, it is just that the default configuration does not support multiple LUNS. The storage array documentation should have described how to set up a Solaris machine by modifying /kernel/drv/sd.conf:
http://newsgroups.derkeiler.com/Archive/Comp/comp.arch.storage/2006…
Solaris is not backward, it different. Just as AIX, HP-UX and Linux are different.
I stopped reading at “The goal of Project Indiana is basically to create GNU/OpenSolaris, slap on a GNOME desktop”. If that’s what you think Project Indiana is all about, then I’m not surprised the rest of the article is equally misguided.
Another thing that drives me crazy about linux is the necessity of recompiling 3rd party drivers for every kernel. I don’t know about *BSD, but it is nice not having to do this in Solaris.
The first problem with this “article” is define “slow”. Since the author chose not to provide any specifics such as sar or system accounting data or even empirical measurements, what is slow? And it is slow compared to what exactly? We don’t even know what kind or size of hard drive or network card is used in the system; all we know is the system in question has a 2 GHz Pentium IV CPU, 2 GB of memory and an nVidia 6200 video card.
So we have an older system with an almost five year old value priced AGP video card, 2 GB of memory and an unknown amount of hard disk space and network card and we are supposed to take the author’s word for it that OpenSolaris and all applications run slowly on it? Just because someone says it is slow does not mean it is. If someone is going to complain about performance, they should back it up with some meaningful data and just complain, because all I have read is complaints without substance.
The author then moves on to the usual OpenSolaris complaints about the lack of software in the repositories. Again no specifics as to what is missing in the way of software. For that matter why doesn’t the author take the time to compile the software he wants himself? I have asked this question before and was shouted down for even suggesting such a thing. For that matter why doesn’t the author or the other people who complain stand up and say “I will contribute to the effort and compile “enter software title here†and put it in the repository� It is easy to complain, but not so easy to do something. How much of a geek are you if you can’t compile your own software?
What user experience? Again no real frame of reference other than mentioning the user experience is better using Fedora, OpenSUSE and Ubuntu.
And this is considered a “quality†article for OSNews? Of course this has not stopped the anti-Sun trolls from piling on. Why was this published in the first place? For a site that is supposed to be dedicated to technology, it is probably the worst piece of so called “tech writing†I have ever read.
You can try it yourself, but you’ll see that a lot of common open source software either won’t compile on (Open)Solaris at all, or it will require a lot of effort to make it compile. Why? Because most of OSS was developed on Linux, which is *not* fully POSIX compliant, even if Linux folks like to claim it is.
OpenSolaris’ main goal insn’t about user experience, Thom just doesn’t (want to) realise that.
Here’s why: http://www.osnews.com/story/21476/A_Note_About_Ads
Because of lower incomes, OS news authors are resorting to very poor flame-bait articles (and other crap). They have zero value, but guarantee a few more ad viewers.
I wrote an article for OSNews on OpenSolaris and used iozone to gather benchmarks on ZFS and did this without incident, the same way I did it with Solaris from 7 through 10. I think the problem for most Linux users trying to do things on OpenSolaris is that they have no idea how Solaris and OpenSolaris works. From simple things like adding /usr/sfw/bin to your PATH statement to using crle to update your dynamic linking environment, if it doesn’t work it is much easier to complain than it is to actually learn how to use it. And just because it doesn’t work exactly like Linux doesn’t cut it for me as an answer. I don’t have the luxury of walking around with blinders and see only “Linux”. If the people I work for want AIX or HP-UX, I can’t say “Well I only know this …” and expect to continue to be employed.
That does not mean I totally disagree with you, I have found several pieces of software that just plain doesn’t compile on Solaris and would probably fail in the same fashion on OpenSolaris like Nagios and OpenLDAP. I’ll have to give it a try on my “slow” (Pentium IV) OpenSolaris machine at home just for laughs.
If the OSNews staff would get some real content and kick the trolls out, I don’t think they would have a money problem.
Just two quick notes:
– crle is deprecated, prefered way to configure locations of libraries is LD_LIBRARY_PATH,
– you can get binary packages of nagions and openldap for (Open)Solaris from blastwave.
Since when was crle deprecated? It is my understanding that using LD_LIBRARY_PATH is bad for security reasons.
Yes I can get Nagios and OpenLDAP from a number of places, I compile applications with particular options that most repositories do not. The version of Nagios on Blastwave is 2.10, the last version I compiled successfully was 3.06. And OpenLDAP has been a pain in rear forever.
Don’t know since when, but I remember reading it somewhere in Sun’s docs. What is the problem with LD_LIBRARY_PATH? I hear (read) this for the first time.
The information in the “Why LD_LIBRARY_PATH is bad” link is old, but it has become acceptable practice among administrators I know:
http://wiki.services.openoffice.org/wiki/LD_LIBRARY_PATH
It also would not entirely surprise me that Sun would potentially drop a feature that is useful.