Today I’m excited to announce our plans to bring SQL Server to Linux as well. This will enable SQL Server to deliver a consistent data platform across Windows Server and Linux, as well as on-premises and cloud. We are bringing the core relational database capabilities to preview today, and are targeting availability in mid-2017.
So this is happening. I feel a little cold all of a sudden.
I’ve used it and was quite happy with it when I was in the windows camp, I’m sure I had gripes I’m not remembering, but it wasn’t bad. However the fact that that it’s was not portable made it a non-starter for many projects that run on linux, so I had to leave it behind.
By locking it up on windows MS is just limiting it’s reach, so I think this makes sense. I doubt very much that MSSQL was attracting new users to windows anyways.
Nope, if they support ADODB, you’re doomed. Their implementation of SQL is not standard, the performance is relatively good, the size is monstrous.
Just use FirebirdSQL or PostgreSQL FTW.
Who’s sql actually conforms to any of the SQL standards? I’m not aware of anyone that does.
*rolling dices*
*checking result*
*frowning eyebrows*
Well, actually…
You’ll find that it’s whose, not who’s.
PostgreSQL is the most SQL compliant database of all. SQLite looks at what PostgreSQL does. And PostgreSQL mentions in their documentation what is and isn’t compliant. And also why they aren’t compliant when they are not.
Well, you’ve just validated the OP’s point. What SQL database is fully compliant with the SQL standard? As far as I know, none of them. PostgreSQL may be the most compliant, but not fully compliant.
Makes no sense, because the standard makes no sense.
There are even parts of the standard no database implements (as specified in the standard)
On the bright side of things. MS actually DO require knowing standard syntax if its available in MS SQL Server for its certification. (At least from 2012 and up).
Official documentation for that certificates also put emphasis on it.
Yeah, nice to hear, but it still depend on the engine you are using. For instance, Jet through Ado have limited instruction set regarding string manipulation inside a query. Simply reversing a string or looking from the end is almost to just impossible.
Yeah Thom. I’m a little cold too. I’m trying to figure if this is a fluke.
M$ didn’t write MSSQL server they licensed it from Sybase and Sybase has been available on Linux Unix for a long time. Surprised it took them this long.
They probably had it working on Linux for years and just now is the time.
I think I will stick with PostgreSQL…. If your writing your own stuff for internal company or private use there is no benefit to running MS SQL server and it’s odd transact SQL. Ya I dont really care for it, PostgreSQL’s MVC control is way better.
The didn’t just “license” it, they had a full partnership with Sybase (sharing of source code) that lasted about 5 years. That ended in 1994 (22 years ago), and the product has been developed solely by Microsoft since then with no involvement from Sybase (now owned by SAP AG).
There have been hundreds of major changes in the product since 1994, none of which exist in the same form in any Sybase product:
complete rewrite of Storage Engine
complete rewrite of Query Processor
complete rewrite of Database Engine
OLE DB/ADO support
OLAP Services
Reporting Services
ETL Services through DTS
CLR integration
XML Support
etc…
The two products share a bit of primordial DNA, but are barely related anymore. Dave Campbell (current CTO at Microsoft) was hired originally to work on MSSQL and spent 10 years on that team. According to him by the time he left the team there wasn’t a single line of code left that wasn’t written by a Microsoft employee – and that was back in 2006…
Just saying, by this point the fact that the product shares common roots with Sybase is no longer relevant in any meaningful way. You can see some evidence of the shared lineage in the basic SQL dialect, but such overlap is purely superficial now.
> According to him by the time he left the team there wasn’t a single line of code left that wasn’t written by a Microsoft employee
If only they could do it with Skype!
I agree they’ve done a lot. But do you really want to highlight ADO support and XML support as part of those large change sets?
ADO is just MS own DB connection protocol, huge surprise or change there…
And XML support is just odd. Who stores XML in a database and why? Thats like a Y2k era bullet point every product had to have.
Its not about storing XML in DB.
Any RDBMS can do that. XML is just plain old text after all.
MS SQL allow XPath to be part of the SELECT Query. Sometimes useful.
I do prefer PostgreSQL JSON though
And 99% of the time a real pain in the side. I have seen horrid sql and XML data dumped and used in a single field, instead of extending the table, which would have made for a much cleaner solution.
snorkel1,
I thought I’d give postgres a shot for a current project. Now that I have some experience with it on a real project, it feels much less refined than I was expecting.
I sorely missed having a usable merge/upsert feature. This finally came to postgresql in 2016, but is still unavailable in the ubuntu LTS repo.
My biggest gripe has to be that postgres views can’t be modified once they are created! This completely throws off my normal development workflow where I start out with skeleton views and fill them out as the project progresses. You can’t do that in postgresql…
http://dba.stackexchange.com/questions/77564/postgresql-drop-column…
http://stackoverflow.com/questions/3593568/how-to-alter-a-view-in-p…
Regardless of the technical reason, for someone like me coming from other databases where views are trivial to change, this is a major regression. I gave up on using postgresql views entirely, for an RDMS this is insane.
I’m still glad that people are working on alternatives like postgresql. It’s got to be on any shortlist of databases to use if you really want to stay away from corporate control. I just wish it were more polished because IMHO it isn’t ready to replace the corporate databases.
To their credit, they are aware of these problems and others, and they encourage users to submit patches to fix them. It really is developed by the community.
Some developers of postgreSQL are employees of EnterpriseDB, who added extension to PostgreSQL to enable it as a drop-in replacement of Oracle database which will tell you a lot of the product.
Edited 2016-03-09 02:45 UTC
allanregistos,
I’ve heard this elsewhere, and having used oracle without complaint (other than the price I suppose), gave me very high expectations for postgresql. Maybe they were too high because now I feel it’s quite rough around the edges. For me, the problems with views are a serious show stopper. But as long as the developers are willing to keep improving it, that’s a good sign that it will be better next time.
So they were the same product in 1993. That was 23 years ago. I’m pretty sure the code base changed since that time.
I think you will find the code has been been rewritten entirely since then – much like every cell in the body is replaced within 7 years (although that ones not strictly true, so it could be the analogy holds in that sense too).
I doubt many Postgres users will be considering a shift. Companies that:
– Are running on Linux infrastructure
– Want a “paid” database (e.g. for support reasons)
suddenly have a good reason not to deploy Oracle.
Not sure about that. Those kinds of customers are going to be risk adverse. All of MS benchmarks, add ons and history are running on windows. Putting it on new operating system that is radically different, is going to introduce some uncertainties. There is no way I do that unless the OS vendor is also okay with it.
So, who’s it going to be? Who is going to stand up amongst the enterprise Linux Distros and support MSSQL as an approved application? SUSE? Ubuntu? RHEL?
Well, Red Hat and Ubuntu were part of the announcement so I am guessing they will be on board.
Actually, I am sure they all will.
Justin
Welp, that makes sense, given the recent Azure anoucements as well. I can see this being used now.
There are other commercial DBs on Linux besides Oracle.
Besides, organizations that deploy DBs on Linux do so because they do not need/require MSSQL server or want to avoid it altogether. The main value proposition of MSSQL is that it is the de facto big DB server for Windows and it being tied to the whole MS dev tools infrastructure. If you remove Windows out of that equation, there really is little MSSQL offers over its competitors.
Edited 2016-03-09 05:14 UTC
Yeah, agreed. However, there will always be legacy platforms that need to be maintained. With the whole Xarimin purchase and this, you could take MS stuff completely off of MS windows. So if you had a no windows policy, this would help quite a bit.
I do have co workers that used to be in the MSSQL space, every time our system does something weird, they –in jest– suggest switching to MSSQL.
I see what you’re saying. But the whole xamarin thing is rather limited, and does not really help move most of the Windows legacy over to linux. In any case, if and organization has had a no-windows policy, MSSQL is not a “legacy” platform for them by definition.
Edited 2016-03-09 19:56 UTC
I admit I’m out of my depth here, I haven’t done anything non trivial in mono, but if you had an app that was C# + MSSQL with little else there. Its feasible without knowing too much more about it that this would help in the transition.
Although, I agree it would be rather limited.
In a lot of the organizations it’s the applications that dictate what database is to be used.
They tend to differentiate their strengths, in order to not spend resources on war, and dedicate those to advancement and optimization.
Hell is frozen indeed.
But I need to wait until April 1st, so the joke will make sense.
But, if true, I hope (fingers crosed) that the licensing will not be an infernal contract.
Edited 2016-03-08 03:00 UTC
This will be great for organizations with Windows SQL Server already in-house who want to extend their reach into Linux without bringing in new DBMSs. However, I can’t imagine that any current Linux shop would choose to purchase SQL Server when free Linux alternatives already exist and are so well-proven and popular. So I’d predict a pretty small marketshare for SQL Server on Linux.
Still, it’s nice to see a new, more aggressive and customer-focused Microsoft at work. Kudos to MSFT.
Just look at Red Hat. We want true openness, but also service to the code, data & exposure.
As for me, welcoming MS again to the global party.
To the ‘STALLMAN’ community, -which I feel part of it as user- just saying to be careful with license ‘back-flush’ and not assigning long term projects until MS proves long term commitment in this path [Every big Corporative endeavor should be balanced with at least two healthy, individually developed efforts]. [Also be careful with the ‘jacks-of-all-trades’ mercantile campaigns (That’s how Apple took his very own ecosystem by the neck)].
Not missing my aluminum hat comment: This event changes noting the path to a ‘Big Brother’, Corporative commanded future. The same individuals that give life to a Nation could become ‘dispensable’.
Where 4/5 parts of the work could be done by inform-ation, automat-ion and robot-ization.
4/5 parts of the population would be dismissed, effectively, as 1/5 is actually dismissed. [Selling ‘hand-made’ crafts to tourists? -and I’m not going to extend to prison industry and traffic of people (people becoming merchandise)-]
On ‘givers’ and ‘takers’.
Engraved on the ‘System of Systems’ [economic] is this stupid ‘seeders’ and ‘leechers’ concept.
From being dismissed to being ‘discharged’ and forgotten only stands an eviction notice.
That young -educated, multilingual- lady crying because couldn’t sell a single one dollar souvenir, at the sunset, and trying no show -with dignity- her face. Sometime, at the Kingdom of the Temples. [Taking decisions]. So proud of her past.
Of my daughter? Of my grand-daughter? Of yours?
You are wrong, neural AI is going to progressively ‘discharge’ traditional code.
And developers, and programmers -most of them- are going to become ‘dispensable’, too.
Shoo shoo shitty ass bot.
http://www.theverge.com/2016/3/9/11186940/google-robotic-arms-neura…
http://techxplore.com/news/2016-03-authentication-path-nec-acoustic…
I’d say this is because the battle is shifting from control of “the desktop” to control of “the cloud”.
When the service is the product – who cares what it runs on?
I suppose that depends. Are all running platforms treated equally by the service provider? Whoever does that, I think, will ultimately win.
Linux is shipping by default with half the cheap appliances people buy and microsoft is actually porting things to it. Sure, there’s caveats there. But it’s hard to not feel like we’re looking at the end of an era. Though much like the last time everything’s changing dramatically so that it can stay exactly the same but with new labels. Doesn’t change how surreal it feels though.
Basically everyone that is using a highend database on Windows is doing that on MS SQL Server, at least the traditional ones and even several NO-SQL ones.
Where Microsoft has most competition is where they cannot compete: Oracle with their all-in-one solution and DB2 on Linux. Performance, stability, security and scalability and features have been the focus the last decade and they have basically caught up or surpassed so they are surely in the game when a server is discussed. But they were immediately out of the game if the server had to be Linux (or non-Windows in general). Another barrier broken
This is very interesting indeed… I want to read the fine print first though…
Yes. It will be interesting to see if the Linux version will have / reach feature parity with the Windows version within a reasonable time (to me, that is about a year), or if it will remain a “starter” version, designed to attract customers, and then force them over on the Windows version when they need more advanced features.
Am I the only one who interprets that as a desperate maneuver as they are losing ground in the server segment?
From what I understand, ASP.NET has pretty much tanked in recent years with the ascent of Node.js. Database servers could be next so they better cover all bases at least.
Not really, ASP.Net has maintened a fairly constant market share since the release of Node.js http://trends.builtwith.com/framework
People can be very vocal Against .Net but its still popular and in wide use.
Microsoft have ported MSSQL as it wants to offer it as a cloud service to non-microsoft users. This effectivly positions them as a direct (and cheaper) competitor to Oracle workloads
Edited 2016-03-08 12:06 UTC
Thanks Adurbe. Also see as ORACLE motivated. On differing from ORACLE, MS stands a slight chance of creating an ecosystem around it.
Come on, who will really do the buy ?
This is almost a joke, or a niche migration project.
– Big companies use Oracle dB on big Irons if dBs are critical for their businesses.
– Companies use Oracle dB on Major Irons if dBs are critical for their businesses.
– Medium Companies use either Oracle, Protgress, MySQL if they find someone to maintain their dB, and if they can afford it.
All the others use MS-SQL or MySQL server because it is cheap.
MS-SQL/Linux is not for low average IT Joe to maintain small business more easily…
… When it is not the manager who wants to put its XLS data in a dB…
If you want to do serious business with databases, you already have a serious solution, or you plan to buy one.
… Not something risky.
You may write whatever you want about this news…
I am sure of one things : It is a flat announce from Microsoft, just to see, and I will never believe it will go to success.
Even the biggest ones need of sound -and quick to replace and implement- plan B (and C).
If you had asked me to put money on which product Microsoft would port first to Linux; this would not have been on the list. Nor could I imagine they would wait this long!
I don’t know, their Azure mobile services are already a node.js front end facade over MS SQL Server, and Azure does run and manage plenty of Linux servers. I don’t see why this isn’t a completely natural extension of everything they’ve been doing over the last couple years – or longer if you look at their cloud strategy (from which CEO Satya Nadella comes).
Any company needs revenue, and Microsoft’s plan to get Windows on to every system has proven unworkable. Since they still want to sell their wares, it makes sense to put it on the platforms people use. #NoBrainer
Just a couple thoughts on this:
1. The likelihood of this being used by people who aren’t actively trying to support software that depends on MS-SQL or have existing infrastructure that uses it is relatively slim.
2. I’m curious to see what Linux distributions they choose to support. If they’re smart, they’ll take a similar approach to Steam and Chrome, and bundle all their dependencies in such a way that they can just be installed in their own directory in /opt or /usr/local, and then shouldn’t need to worry too much about distributions (Steam for example only officially supports Ubuntu, but it’s relatively easy to get it working on almost any Linux system because of their packaging). I think it’s more likely though that they’ll stick to RHEL, SLES, and possibly OEL support, and make it in such a way that you need a dedicated server for any kind of isolation.
3. I really hope that some of the bigger companies that depend on MS-SQL make a similar move with their products. One of the last things where I work tying us to Windows servers is AutoDesk Vault, which has a hard dependency on MS-SQL for metadata storage. I would absolutely love to see the server component for Vault be able to run on Linux (even if it meant we had to use CentOS).
Good thoughts. On #2, what MS Should really do is just ship a docker container as their product. Screw anything else.
#2. They will support Ubuntu first, then will rely on feedback of what distro to support next and supporting other distro is 1+1.
And I missed the apocalypse.
You are seeing a mirage on the horizon. Please take video to show on your coming back from vacation
…when MS ports Office to desktop Linux, then Hell will have frozen over.
There is Word and Excel online as free Chrome app on Linux.
Can’t get Chrome to work on Ubuntu 14.04 LTS, though. Something about outdated GTK packages.
And now we know why the year of the Linux desktop never happened, nor ever will.
Yup. Fragmentation, no stable ABI.
Uh oh, you said the three-letter curse word.
You all know this was -and is- just a little will problem. Doesn’t it?
That’s odd. I have Chrome and Chromium running on 14.04 no problem.
Or your own install?
Half-baked and slow compared to the real thing.
And will pay for servicing MY code and MY data and MY exposure. [I’ll manage!, thanks
]
Not believing it will happen on my lifetime.
The modularization [along false protocollary surfaces] and open-ization ‘show’ -and the consequent bulling of the ISO committee- convinced nobody.
Should say that dominant Offices at Linux land still very lacking on modularization.
Thats the problem with Linux is there’s no standard ABI between different distributions.
I’m so sick of this problem with the stuff we write, which we produce binaries. So, what we do is develop on RHEL, which guarantees a stable ABI, and test on a few others, if it works, great, if not, sorry, we don’t support that distro.
RHEL and SuSE enterprise have pretty decent compatibility between them. The rest, forget it.
MacMan,
Usually that’s said with regards to the linux kernel code, where I agree having no stable ABI can be a problem. But I’ve never encountered a linux ABI problem with userspace apps. The userspace syscalls are exceptionally stable even across distros, this is why you can compile a vanilla kernel yourself and it should work in almost any distribution without any customization – because there’s hardly any ABI changes for the userspace interfaces. If you have a specific ABI incompatibility in mind, I’d like to know about it.
Do you mean different userspace dependencies? If so, then yes that makes sense (technically they aren’t part of “linux”). I agree shared libraries can be very different between distros. There’s no reason you can’t bundle all the dependencies you’ll need on a different distro, but it’s a bit like “DLL Hell”.
In the particular case of porting the MSSQL database, my guess is that MS depends on very few third party library dependencies (apart from the most basic ones like stdc).
Edited 2016-03-08 21:09 UTC
Both, if wanting a foothold in Europe. A kind of LSB should be brought back from ‘zombie’ land. This time on a real participative modus operandi. Please avoid large-blocks|oligopolic-mantainers.
Which means your product must be used all of the known 300 Linux distributions? What kind of product you develop since you are so concern of incompatibilities between distribution? Incompatibilities within distribution rarely will affect user space applications. And even if it so, you only need to support the big three = RHEL/CentOS, Suse and Debian/Ubuntu. If your users used obscure distributions, then that is not your fault.
Find FCC’s Tom plan more centered and resilient.
http://phys.org/news/2016-03-unveils-low-income-broadband-subsidy.h…