Restaurant booked, flowers ordered ... Microsoft has a hot date for SQL Server 2016 Microsoft has announced that SQL Server 2016 for Windows will go into general availability on June 1 and is positioning it against Oracle on cost, security, and speed. The database management suite will come in three flavors – Enterprise, Standard, and Express – as well as a Developer SKU. Customers with a Microsoft Software …
Wake up, it's not 1998 any more.
SQL is a good functional database platform.
I'd take SQL over Oracle's price gouging and ridiculous toolset. Not to mention the abortion that is RAC. I can't think of many reasons that enterprises would persist in subsidising Larry's latest gin palace.
Not to mention the Jet Red database engine that is traditionally used in Access is totally and completely different from SQL Server.
Why they haven't moved Access and the like over to SQL Server express LocalDB (the tiny version that links as a library and runs with about a 2MB RAM footprint) or SQL Server Compact yet I'm not entirely sure.
Nobody should really be using Jet Red these days, it's just not that good at what it does.
answer:
Systems that only run on SQLServer.
There are enough companies out there that still believe that the MS Stack
- MSMQ
- .Net
- SQLServer
- Biztalk
Is the best thing since sliced bread. Sadly I work for one of them. Of the things listed above SQLServer is the least of our worries. MSMQ and Biztalk are the real dogs.
Looks like Postgres is going after a slightly different market or markets. SQL Server tends to do well in vertically integrated stacks using lots of Windows tools.
Postgres via thinks like Enterprise DB's Postgres Plus is targeting smaller Oracle shops with the promise of minimal migration pain and much, much lower licensing and support costs.
The most interesting stuff seems to be going on the proliferation of different backends / storages which make Postgres interesting for the "big data" crowd: time series, columnar storage, etc.
>Unfortunately, PostGres is most similar to Oracle, so in a lot of cases isn't what people currently using SQL Server are after.
Not sure I understand, I work for a company that supports a number of RDBMS' as underlying database and I do not understand why any vendor would want to limit you to a specific RDBMS and how these vendors find customers ... you have to be pretty dumb to use a limited toolset in that respect.
Sybase is very similar to MS SQL Server, and has managed to get in-memory databases working properly, as I have already written, Sybase ASE beats MS SQL Server hands down (in our use-case, at least) - and I work for neither SAP nor Sybase. I am really looking forward to testing MS SQL Server 2016 GA, I will let you know the outcome of our tests.
And please, Windows Cleaner brigade, please stop referring to MS SQL Server as an enterprise DB, it is not. As soon as the dataset-size increases, it grinds to a halt ... where PostgreSQL continues to run along quite happily and neither Oracle or DB2 seem affected, at all.
As for those claiming PostgreSQL is free, so is MS SQL Server, DB2, and Oracle ... the latter two without enterprise features.
In most cases, the only reason why you are using MS SQL Server is: "it runs on a OS I know how to use with my mouse and query analyzer allows me to drag-drop queries together which helps a lot, because I can only type with two fingers and I freak out when I see the old-school black boxes with flashing cursors and scrolling text ... "
"If you do not know either DB2 or Oracle, your opinion on RDBMS' simply does not count!"™
"when PostgreSQL is faster"
Do fee free to post some recognised benchmarks to back that up? In all recent examples I have seen, SQL Server was much faster.
"better"
How so? Vastly smaller feature set, much less support from other products, and more security vulnerabilities don't exactly seem like that to me...
"and available for free?"
If your time, cost of additional hardware for required similar performance, and security and system capabilities have no value to you - and you don't need support?
"So you request benchmarks yet provide anecdotal evidence? Remind me not to hire you for advice."
An example of the many similar comments I see, and of my experience:
http://www.postgresql.org/message-id/60zlrwuyqb.fsf@dba2.int.libertyrms.com
Postgres is generally way way slower than MS SQL server and whilst SQL Server has no difficulties at all handling vast databases with 100s of TB, Postgres runs into multiple scalability issues at anything over modest sizing.
@TheVogon
I am surprised you posted that, you usually READ what you source, and I think that in this case, you did not read the reply from the postgreSQL guy, the devil is in the detail. Especially, the part about "locking" ... that will bite you when you use MS SQL Server.
As for your joy, suck on this:
http://www.pg-versus-ms.com/
Ouch, I know ... I am sorry for you, TheVogon, yet again, but this time, you even provided me with the stick, thanks!
"http://www.pg-versus-ms.com/"
That page supports what I said about Postgres performance sucking. In between lots of largely uninformed rants that are basically moans that MS SQL does certain things in a different way that he clearly doesn't understand - for instance "I don't know how easy MS SQL Server is to update".
They were the same thing in what, 1991 or so? Today if you know one you'll fell at home on the other quite quickly but they've both been re-written several times since 1991!
They're both good RDBMSs and each have features the other doesn't but each new version both closes and opens the gap. Sybase is now called SAP ASE of course.
I work with both a lot and personally I slightly prefer ASE but I prefer either over Oracle any day of the week.
1991 ? You must be kidding, right, more like 2001! Look it up, ever heard of Wikipedia ?
https://en.wikipedia.org/wiki/Microsoft_SQL_Server#Genesis
"SQL Server 6.5 was release in 1996."
"SQL Server 7.0 [released in 1998] and SQL Server 2000 [released in, you guessed it, 2000] included modifications and extensions to the Sybase code base, adding support for the IA-64 architecture. By SQL Server 2005 the legacy Sybase code had been completely rewritten."
Now, both DB's speak TSQL, MS has added quite a lot of extensions to TSQL, so has Sybase, both after the split, but, and this is a BIG but, Sybase has managed to get in-memory tables right, years before MS thought of it. So they have the lead there ... if pure performance on small to medium datasets is your requirement, SAP ASE (or whatever it is called now) makes much more sense than SQL Server - if you have TSQL knowledge in-house, that is. MS SQL Server sucks golf balls through garden hoses on large datasets, as far as i have seen, YMMV.
PostgreSQL makes much more sense on medium to large datasets if you cannot afford Oracle or DB2 ... again, as far as I have seen, YMMV.
Now, if MS SQL Server makes little sense on small, medium, and large datasets, the question is, where does it make sense ? At least they have improved performance, I hope that is on medium to large datasets ... I also hope, for them, that they finally got in-memory tables right, but I doubt that ... ;-)
The common code does not really matter ... MS SQL Server has "backwards" compatibility, of course, so it is relatively easy to migrate from MS SQL Server to Sybase, more so than to move to PostgreSQL or MariaDB, for example.
The thing is, of course, that the extensions that MS did to the TSQL will have to be "converted" to how Sybase extended TSQL, if you use that ... and some things will need to be adapted here and there, however, all the built-in stored procedures, for example, are there in both.
Ouch, once more, you did not read .... can you read? He was linking to the express edition, as in, the free of charge edition of SQL Server. I do assume this is slightly different for the commercial versions ... then again, if anybody can link to a story anywhere where MS paid out to a customer because of a fuck-up on MS' part, any MS software counts, I would like to know ... I am pretty sure that you will not be able to.
So now, if SQL Server 2016 is boasting up to 34x performance boost, we should be seeing up to 1020x performance boost on SQL Server 2012. That should be fairly easy to spot, your struggling query that was just hitting that absurd default .NET timeout of 30 seconds is now near instantaneous.
Thing is I didn't ever observe that 30x performance boost for 2014, I assume whatever tech they had ready for it got pulled, and we're going to see it in 2016 instead. That's the nicest way I can spin those figures, a cynic might just say that we've heard all this before.
I'm not your downvoter but I did see a huge performance increases of that order moving to SSAS tabular model from SSAS MOLAP.
There's a whole load more in-memory stuff in 2016 that might give similar improvement *if* your workload fits.
Most of the clever new stuff is only in the Enterprise Edition. Over here in the real world of SQL Standard, nothing much has changed since 2012.
Still can't do an indexed view with an outer join either - Oracle manages it, why now SQL?
> cost, security, and speed.
ROFL, what about ... availability ? ROFL ... and, as I wrote so many times in the past, AND MS has acknowledged, " queries run 34x faster", I have not tested, but I guess MS SQL Server now matches Oracle 11, not bad, really!
Now, if they could get cluster support on Linux, we might get somewhere ... MS Window's TCP/IP stack is soooo sloooooow, unless you have SQL Server on the same box, you are wasting resources left, right, and center ... at least it can accomodate more than 50 concurrent users and thus beats MS Access ... ROFL
How do the "Windows Cleaner and Surface Specialists" feel today ?
> if they could get cluster support on Linux, we might get somewhere
The SQL Server clustering model is someway ahead of anything I'm currently aware of on Linux outside of a few very expensive niche commercial options such as Oracle RAC, so I doubt that will be possible. SQL Server on Linux option is most likely just a sop to developers who want the power of SQL Server in custom test / dev environments - if you want the full feature set and performance it's going to need Windows Server.
> MS Window's TCP/IP stack is soooo sloooooow
Windows Server outperformed Linux in my IP testing of 40 gbit adaptors (higher throughput @ lower CPU load) and also had a higher top speed as an NFS server! Also you have ultra low latency methods on Windows like SMB Direct that as far as I know are not available on Linux.
So I would be interested in any actual empirical evidence suggesting performance restrictions on recent versions of the Windows Server IP stack?
>"Right now they're like a second-best footballer who only uses one leg. With the Linux port, Microsoft gets access to a substantial business they didn't have before."
Come on, the only RDBMS I know that is worse than SQL Server is MS Access (Ok, does not really count, so MS SQL Server is the worst), performance wise (on same hardware) and I only know SQL Server, Sybase, MySQL, PostgreSQL, Oracle, DB2, and MaxDB ...
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
