Submitted by Scale-with-rails compares Solaris Express to FreeBSD and then goes on to demonstrate the power of the Sun Niagara based T2000.
FreeBSD vs. Solaris Express
About The Author
Eugenia Loli
Ex-programmer, ex-editor in chief at OSNews.com, now a visual artist/filmmaker.
Follow me on Twitter @EugeniaLoli
As much as I like Solaris, that powerpoint really went on to show nothing other than a simple netperf test. Whoopity-doo. The “attitude” displayed in the PP was pretty lame imo as well, I am sure the author intended it to be “fun” or “lively” but it came off as ignorant and half-assed. The grey hair picture was the only funny part of the slideshow. The “FSCK” slide really showed the lack of thought/common sense put into that presentation. I’d love to see the reaction of any non-super-nerd/non-immature person that got past the “wee is funny” stage. (Pun intended..)
That being said, it sounds like that group has done well, and I applaud them for that. Congrats.
Edited 2006-04-28 01:25
I agree completely.
When I started to read this PP it made me laugh at first, then it pretty much just lost me as it went on… I found it to not focus at all with the title it was given. I expected a few more tests with a lot more information.
It also gave me this strange Ubuntu feeling with the brown background… totally off topic of course, just my opinion.
As for Solaris vs FreeBSD I have always had better results using FreeBSD. However Solaris is very good, just not my first choice. Remember, speed isn’t everything. I’d rather go with the most consistent performance over the “can be faster here” OS. Though this is again slightly off-topic.
Remember, speed isn’t everything. I’d rather go with the most consistent performance over the “can be faster here”
A M E N !
sorry to be pedantic, but it was a PDF, not a Powerpoint file.
So the link on the page is to a blog that has a link to a pdf that looks like a powerpoint presentation. There is some good info but it’s hard to grasp in a powerpoint presentation. It looks like Solaris beat the crap out of FreeBSD but I would like to see a similar comparison with Linux, considering it is more widely used.
There was another test posted a few weeks ago that compared Solaris with Linux. And showed Linux to be a bit faster, obviously.
http://www.anandtech.com/IT/showdoc.aspx?i=2727&p=7
Can we please not link directly to blogs which link directly to whatever the “source” is? I can understand if the blog has some useful information, but this one just says (paraphrased) “go watch this powerpoint, it says Solaris is cool oh and btw there are tons of other cool things like *insert names of features in Solaris*”.
The blog added *0* to the content. Just link us to the powerpoint, with a disclaimer that it’s a PP. I think it’s pretty bad form to even link to a PP with no text-based translation of the content in the first place. Not everybody has PP.
> Can we please not link directly to blogs which link directly to whatever the “source” is?
I agree with you on this one.
BTW, a fun presentation.
I had a good laugh on the “fsck” slide (which makes me “super-nerd/immature”).
For that slide only:
http://www.flickr.com/photos/textdriveinc/133831135/
Super-nerd wasn’t meant as offensive.
I was just trying to point out that PP wouldn’t be impressing the general public/business folk. I suppose that wasn’t their aim though. To each their own!
Hmmm.
Wouldn’t perhaps “fsck -F you” or “fsck_you” be funnier?
Alright, who wants to write the “you” filesystem for Solaris?
I’ve seen various flattering benchmarks, but how has take-up been in the real world? Anyone consider it and then reject/accept it? I’ve been watching this thing for years and am terribly curious about it.
I’m not sure if this is perfectly the domain of OSNews but at least its related.
Based on the limited testing I did on the T2000 I had, the machine definitely lives up to the hype.
Now I’ve never jumped to defend FreeBSD before, but
the FreeBSD “comparison” is useless.
There is real no indication of platform, network
devices, configurations, versions of the OSes, the
network itself.
Such a huge disparity in performance suggests to me
that either something is misconfigured, or there is
a driver bug; or that the test is being performed on
a reasonable sized SMP system, which FreeBSD has had
trouble with.
There is plenty of evidence to suggest FreeBSD should
have no problem filling a gige pipe with a single
decent CPU (though I haven’t tested this myself).
I realize Solaris may be the way to go (assuming it truly is faster and more capable than FreeBSD), but I also worry about Sun’s ability to stay alive in the coming years. I would much rather trust my business to FreeBSD, with all of its faults, then to Sun, only to have them go out of business and leave me high and dry.
I know I can count on FreeBSD to be there tomorrow and the next day, and I also know the FreeBSD team is constantly improving the operating system.
Sometimes the desire to use a “free” operating system has nothing to do with price or politics. Sometimes it has everything to do with knowing you won’t have to find another solution tomorrow, after your current “wonder OS” has closed up shop.
Solaris is open source now.
If Sun die, community will continue to improve Solaris 10.
> I realize Solaris may be the way to go (assuming it truly is faster and more capable than FreeBSD), but I also worry about Sun’s ability to stay alive in the coming years. I would much rather trust my business to FreeBSD, with all of its faults, then to Sun, only to have them go out of business and leave me high and dry.
Well, as an indicator keeping in mind that more than 50% of all Oracle deployments are taking place on Solaris, I don’t think you should have any worries with the future of Solaris. If anything Solaris is growing and will grow even faster in the future. It looks like people are finally starting to get the message on how technologically good Solaris is, OpenSolaris is the fastest growing open source OS, so future should be quite bright for Solaris.
can someone post the pdf link, my work bans blogspot.com
http://www.scalewithrails.com/downloads/ScaleWithRails-April2006.pd…
Shouldn’t you be working anyway?
http://www.scalewithrails.com/downloads/ScaleWithRails-April2006.pd…
I manage more than 30 FreeBSD servers and manage to fully saturate a gigabit interface during rsync backups everyday.
Have no idea where or how this guy is getting his results. I notice he’s pretty stark regarding comments on the details of his testing.
I have brought up the point various times in the past here about a total lack of information as to the specifics of a test and got flamed for it. This is another example of comparing oranges to something, but you have no idea what that something is.
I guess it is more important to push an agenda by any means available than to publish something that can be verified or duplicated by a third party.
This is just basically saying that Solaris runs better on a heavily threaded CPU. When was this news? Let’s wait until FreeBSD is fully finished with the Giant Lock and see the results then. Until then, noone would even consider using FreeBSD on a Niagra Chipset. Let’s see how FreeBSD and Solaris compare on an AMD chip.
Was that the test bed? Both Sun and FreeBSD tested on Sun hardware?
Well that’s hardly fair is it?
Actually it appears the network performance test was
comparing FreeBSD and Solaris on the same single CPU,
single core system. See slide 40 for the results.
For a bit more detail, see
http://textsnippets.com/posts/show/399
But this isn’t just about network performance. It’s
also about idle CPU, the use of lots of memory (slide
57). That is one of the reason they discuss the
advantages of using the T1000 and zones (slide 59)
along with rack space and cooling (see slides 67 & 69).
The presentation was done by one of the main guys at Joyent/TextDrive (the company that does a lot of Rails hosting). The performance tests referenced can be seen seen on the TextDrive weblog (http://weblog.textdrive.com/).
As for the “fsck you”, I believe it was a reference to DHH’s earlier slideshow at Canada on Rails (http://www.flickr.com/photos/planetargon/127984254/). From what I understand, it prompted several other presenters to be a bit more daring.