SQL Server 2008 - from semi-relational to sublime
SQL Server 2008 R2 is a step closer to reality. On the heels of August's first code drop, Microsoft has released a second, more-fully-featured community technology preview (CTP) of its next database server. It promises a number of things, including improved business intelligence through database changes and integration with …
otherwise people will think it's obsolete on release...
more than a review :-)
Yes, it's true that Microsoft has improved a great deal, but I think I still prefer ASE and Sybase IQ. IQ in particular is streets ahead.
Given that, I'd probably take MS SQL Server ahead of Oracle, if only they'd do a UNIX/Linux version!
A micros~1 fanboi!
I think I'll stick to my more practical and less expensive religion, open source.
Yes, sure, but does it work?
The last time I messed with MS-SQL it went into a "log-file full" mode and eventually refused to work.
So having managed to be not so far behind real databases is an achievement?
Technical savvy people would need to know the extensions of the files on their machines and would also have no problem in finding the configuration and changing it.
Hiding the extension is a "feature" aimed at the average Windows user - presumably to avoid overloading their little brains with information such as the extension of a file and - GASP - what a file extension is.
Oh, you so clever, because you take thing from what they used to do and make it a joke, yeah! And then you use IT angle avatar, because Microsoft is not IT. Oh, so clever joke boy you!
MS SQL is a quality product, not my first choice in database but lightyears ahead of anything the free and open source community has put out in the area. Unless Google releases BigTable I don't see that changing any time soon, either. But hey, it's easier to dismiss a Microsoft database slated for release in 2010 based on the fact that you don't approve of an at the time reasonable limitation they imposed on filename length in 1981 than have an independent thought. :/
"... three source systems holding customer lists but none of them has a complete set. MDS makes it much easier to consolidate the three lists into one complete, accurate list of customers ... it allows the data to be synchronised across the three sources so that each has the same version of the data."
That sounds vaguely familiar. Oh yeah - Lotus Notes/Domino has been doing it since 1989 when v1 was released. It's called "replication". Yawn ...
"MS SQL is a quality product, not my first choice in database but lightyears ahead of anything the free and open source community has put out in the area."
What? Never heard of PostgreSQL then? Take a look at any comprehensive comparison of RDBMS packages and you'll discover that not only does PostgreSQL match what SQL Server can offer but it's approaching Oracle (and has already overtaken Oracle in price-performance). Not only does PostgreSQL have rock-solid stability (I've been running data-bases at several sites with each new version of PostgreSQL under Linux for seven years without a single software failure), but it comes without all the licensing baggage of MS software.
This is a content free marketing rehash of no use to the serious practitioner.
SQLServer still has a clinically dead optimizer, aimed at trivial data base sizes. It relies blindly on the internal statistics, despite the fact that statistics collection is based on a terminally broken algorithm conceived by someone (probably decades ago) and never adjusted (and not user adjustable). SQLServer will happily decide to read 1 billion rows (or a trillion) rather than do a direct index key lookup to retrieve a trivial result set when all values required are specified in the predicate. It takes a seriously warped engineering attitude to think that this is appropriate under any circumstances.
SQLServer is also a failure with respect to adherence to standards, never mind what might be considered useful. The inability to describe complex contstraints declaritively springs to mind and the necessity to drop computed columns in order to change underlying functions is another example of what a dirt cheap implementation SQLServer actually is. Domains I hear you ask? Nope, never heard of them, and don't try using user defined types, or you will discover that they don't really work, and woe betide the poor fool who might want to alter a column so constrained.
The number of View declarations and complex select statements that cannot be correctly resolved, or simply return the wrong answer is horrifying. Nothing has improved in the SQLServer2000 to SQLServer2008sp1 timeframe in this area.
I could fill up a book with all the things that are fundamentally broken with SQLServer. However, MS is not and has rarely been interested in building a solid underlying SQL engine. They are interested in seducing incompetent journalist, fanbois and diverse lackys with an alphabet soup of this months flavoured functionality. And today I read yet another version of how successful MS are at actually getting their message out.
Now, I know that there are some great engineers in the SQLServer team, like whoever did the page compression code, which is an outstanding piece of engineering with which I have yet to find fault. But the bottom line is, that MS is still concerned with the alphabet soup, flavour of the month than core engine functionality - the stuff that makes DBA's lives somewhat tolerable is sadly left far behind. Gee, if only the basic functionality did what it says on the box. Don't get me started on replication - another disaster on wheels that MS still cannot get right. SSMS is also seriously creaky and did not improve in the last iteration - actually, now I think of it it got provably worse as they removed functionality that worked, and replaced it with functionality that didn't work. Multiple bugfixes and the issue remains - but hey, no one uses schemas do they? So why bother fixing it?
Enough of this rant. Whatever else should be obvious is that the issues raised above are probably not even on the radar of MS, the MS marketing machine, the reviewer or the rest of the SQLServer fanbois crowd, almost none of which could manage a piss-up in a brewery if the chatter on the internet by self styled gurus and experts is anything to guage by. And that fact really says everything about what MS is as a corporation and has always been.
SQLServer2008R2 is just the next iteration of more new features at the cost of actually having base functionality that actually does what it is supposed to.
Dr. Dweeb
Yay, a well formed put-down of MS SQL that isn't "Aye Hates It".
As for the author of this piece of advertising, I realise that saying something like "you should be ashamed" will have no effect, since it's obvious that, as a marketing/advertising droid, shame has already been surgically extracted. However, El Reg should be ashamed for tagging this as a "review" when it's nothing more than a thinly veiled rewrite of a marketing press release. Did the author perhaps get a free "gift" in exchange for this fluff? I guess we'll never know.
I had to pinch myself reading this "review"... Now I use SQL Server (2000,2005,2008) on a daily basis (amongst others, but let's not start a crusade here), and I am no MS basher but seriously this has to be the most childish naive un-technical evangelistic "review" that graced el Reg in a bloody long time, and I do hope that it takes as long (or longer) before another one appears because this kinda smells bad, even for a vulture.
Microsoft always seem to be several years behind Oracle and SAP on this. Obviously those two companies bought their way in (Hyperion, Business Objects) but the technology is more advanced than Microsoft.
All this talk of SQL Server being "fundamentally broken", "several years behind", failing in its "adherence to standards" (WTF?), etc, etc, is pathetic and irrational.
My own experience of SQL Server is that it runs our business (a web-based application) just fine, tens of thousands of users, sophisticated integration, no complaints about performance, no lack of features we've needed thus far, and no downtime (attributable to MSSQL) in 2.5 years.
I smell incompetence and blind prejudice; in other words, bullshitters.
That said, this article wasn't great!
Are you suggesting I am incompetent?
Now that is really funny - rotfl.
Dr. Dweeb
"The last time I messed with MS-SQL it went into a "log-file full" mode and eventually refused to work."
Given that there are a lot of MS SQL DBs out there that work fine, Occam's Razor tells me that the fault isn't with MS SQL, there. It's with you.
Blah blah Microsoft blah blah hate, blah blah open source blah blah superior. Blah blah Oracle, blah blah DB2, blah blah Postgres.
Your post is a great critique of SQL Server and I honestly stand behind you on everything you say - I must have encountered most of the things you pick out at the most inconvenient moments myself.
However, will that stop me betting my mortage (in a very real sense) on the continuing success of the Microsoft product? Not a chance. Long live the marketing dollar.
"The last time I messed with MS-SQL it went into a "log-file full" mode and eventually refused to work."
Depending on the recovery model selected, it will write all modification operations into the transaction log. If you don't back up the transaction log to free things up, it will become full - this is all in the online help. Sounds like an incompetent DBA (not doing regular backups is bad), or a (system?) DB got hit with a lot of updates and had a small log size set.
