I'm investigating the possibility of upgrading our SQL Server from SQL Server 2000 to either 2005 or 2008. There's a lot of concern over whether or not 2008 is really ready for us to use because the database is so critical to our product. What are the compelling reasons to use 2008 over 2005? Have you had good or bad experiences with 2008? I'm particurally interested in query performance, new features for our queries, and stability.
One good reason--Backup Compression in SQL 2008. SQL 2008 has 3 options for backups:
In SQL 2008 there are 3 options while backing up databases, these are:
No Compression (same as SQL 2005)
Server Setting (if server backup
compression setting is on/off - use
this setting)
Compression (Compress backup file
during backup)
SQL 2008 backup with compression is generally faster than SQL 2005 backup without compression (less file i/o with a compressed file)
Edit: We've been using SQL 2008 for a while now and have had no problems. In fact, working with the new features in SQL Management Studio (such as intellisense) have made some tasks easier.
AS for new features, I have to point out the new spatial datatypes, geometry and geography, which basically allow you to add actual longitude / latitude points and even cooler, bind those points to a geometry.
i.e. imagine a map with zipcodes, the geography representing the center of the zipcode, and the geometry the are the zipcode covers. Then you could use the built in new function of sql server to determine if a customer for instance is in that are.
Edit P.S. here's a link to a series of articles describing benefits / new features:
What’s New in SQL Server 2008 Part 3
I'd upgrade to 2008 because:
While you don't think about it now, 2005 will go "end of life" sooner and another upgrade could be forced before you are ready.
Everything 2005 can do, 2008 can do. But the opposite is not true. The worst thing to happen would be to upgrade to 2005 only to discover you need something with 2008 and have to upgrade again.
There are many little things improved or added in 2008.
I'd also ask why are you worried about going to 2008? If you are upgrading today, why go to a product that is 4 years old instead of going to the latest?
We have been using it for about 3 months now, and it seems pretty stable. We also upgraded from 2000 and it was pretty painless. One thing I noticed was the new MERGE statement, which is a nice way to do upserts.
I've poked at 2008 a bit, but haven't used it "for real" yet. I can only offer some highlights.
It is "ready for use". SQL Server has been a totally viable product at least since 7.0. If it was buggy, crappy, or generally worthless, its competitors (everyone who isn't Microsoft) would have made it pretty clear to everyone by now.
It's a new version, it will be supported by MS longer.
Many new and fancy bells and whistles. However, they are bells and whistles, they don't really add that much to "core" database functionality.
There are some serious additions, in particular the various new date and time datatypes. (And if they every get something like Oracle's grid computing, then I'll be really impressed.)
By and large, I believe its "you buy the 2008 license, you get to use 2005 or 2008, your choice". As such, you're better off using the latest version.
Related
I have looked through many pages trying to find this information but all I get is the backbone structure and how the SQL server runs. What I would like to know is the difference of commands. Which commands can you use in 2008 but are not supported in 2005 with or without adjusting the compatibility level. I design different queries and my clients differ in their database versions. So when I create something, I would like to know if it will work before I put the time in or start actually breaking my head over why the code doesn't work on the clients computer but it does on mine. An example would be Pivot where I have to change the compatibility.
So what commands have changed? Either a link to a page or if it's not much to type out, it would be greatly appreciated.
This technet article has a list of the new commands in SQL 2008 T-SQL
There is some info here but it's not a whole lot so I am suspect that it's complete listing
http://msdn.microsoft.com/en-us/library/ms143179(v=sql.100).aspx
Version control!
Usually lower version's commands are available and work well in higher versions. But not quite the other way around. Take a look at the following post.
Advantages of SQL Server 2008 over SQL Server 2005?
For each major version Microsoft does release improvements done or changes made under product info.
Closed. This question is opinion-based. It is not currently accepting answers.
Want to improve this question? Update the question so it can be answered with facts and citations by editing this post.
Closed 8 years ago.
Improve this question
I am rewriting an application for my company which requires a database. We are currently using SQL Server 8 and I will be updating the database. My question is which database to use. I have been looking at both SQL Server 2008 r2 and PostgreSQL 9.
The database will be hosted on a Windows Server 2008 box and will be used for both the stand-alone app and an associated website written in .NET.
Some of my points of concern are:
Cost (obviously PostgreSQL wins out here).
Is it ready for primetime? I have read somethings about it being as good as SQL but would like some input from those using it.
Ease of use. I am very comfortable with SQL Server. I have been experimenting with PostgreSQL and while it is different than SQL Server I am getting the hang of it.
Speed. Is PostgreSQL as fast as SQL Server?
Can I use ADO.NET (.NET 4.0) with PostgreSQL and get the same functionality as Using SQL Server (will be using Linq or Entity Framework)?
I will NOT be converting our old database in SQL Server 8 to the new database so I am not concerned with any conversion.
I would be interested in peoples opinion of the two and if PostgreSQL is an acceptable solution for a business application given the above.
TIA
Brian
Necromancing here.
As pointed out by SteveG, SQL-Server Express is probably what you want.
But I've used SQL-Server, SQL-Server Express and PostgreSQL extensively, and you should be aware that SQL-Server Express has some more nasty limitations other than just the DB size:
Installation requires administrator privileges (and system restart)
ReportingService Express can only connect to a local database,
ReportingService Express cannot use other data-sources than the local sql server (no ODBC)
Some ReportingServices features are missing (e.g. graphs and CSV/XML export formats)
Constrained to the usage of 1GB RAM (Same in 2008/2012)
Constrained to a single CPU (newspeak: Limited to the lesser of 1 Socket or 4 cores)
no Table and index partitioning
no indexed views
no database mirroring
no database tuning advisor
no notification services
cannot send email (you might be able to do it using CLR-Stored procedures if you absolutely need to)
cannot use/call Web-Services (newspeak: no HTTP Endpoints)
no service broker
no fulltext and no semantic search (requires Express Advanced Services)
Profiler tool is not included with SQL Server Express editions. (there is SQL Express profiler on github)
Job Scheduling service is not available with SQL Server Express.
Data import and export feature is not available with SQL Server Express (can/must be downloaded)
does not come with SSIS
and as said
- 4GB database size (raised to 10GB in SQL 2008 R2 and SQL 2012)
Note that the limit of 10GB is neither 10GB nor 10 GiB (10 * 10243), it's actually 10'240 MB (10 * 1'024 MB), which is actually only 95% of the 10 GB that I'd have interpreted as 10GiB (10'240'000'000*100/10'737'418'240)
Add to this the limitation that Microsoft-SQL-Server will only ever run on Windows (to date, Linux-port of SQL-server is now coming, but only as CTP so far, and without BI).
So to the other points often overlooked:
In Windows 7, you can only have up to 20 concurrent connections but no more.
If you need more, you'll have to put your SQL Server installation onto a Windows Server OS.
That means if you use SQL-Server Express in a use case with more than 20 concurrent users, you'll still need a windows server license (the number of connections is not related to SQL Server edition, but to the operating system). You'll probably hit the 1GB RAM barrier before as bottleneck.
If you want to economize, you'll have to run the db engine on the same server as the web server, with all resulting performance penalties.
So now to your points (in order of importance)
5. Yes, there is Npgsql.dll (MIT license), so there is no need for ODBC
BUT: If your code uses System.Data.SqlClient everywhere (e.g. parameters, especially when copy-pasted all over the code) then you will have to re-write more than just the SQL code (Use the abstract classes in System.Data.Common instead, or or the System.Data.IDB* interfaces).
If your code copy-pastes SqlClient everywhere, it could be argued that it needs rewriting anyway.
Ease of use:
A little more of a learning curve than SQL-Server, but overall, not too different.
pgAdmin may not fullfill your requirements if you're used to SSMS, but there are several other management & development tools you can use (e.g. Navicat) that come very close to SSMS or even top it.
4.Speed. Is PostgreSQL as fast as SQL Server?
Yes and no. It can actually be faster. Granted, because SQL-Server has had much work done on the query optimizer, PostgreSQL might be a little slower when compared to the Enterprise Ultimate version of SQL-Server (or whatever the current name of the killer-edition is).
But, who ever has that version ? If you can use all your CPUs with all cores and all the RAM you want with PostgreSQL, you will always be faster than some cut-down Express or Standard version of SQL-Server with 1 CPU and 1 to a few GB of RAM (windows limits you to 64GB in 64-Bit afaik, with Linux you can exceed the 4 GB RAM per application limit even on a 32 bit processor with the bigmem kernel and can get it to address 64 GB RAM on a 32 bit processor, not to mention the 64-Bit limits), no matter how super-over-optimized the query optimizer is on SQL-Server. That said, my personal opinion from using both pg and ms is that PostgreSql has the better query optimizer, but that's just my opinion, I have no data to back that up. Besides, windows (Windows 7) is a slow operating system (again personal opinion, no data to back it up). So when you can let PG run on Linux in headless mode, it will almost certainly be faster than SQL-Server, even the enterprise ultimate edition.
Memory footprint
You can totally configure PostgreSQL. You are in charge of the configuration file.
Can be small or large depending on how you adjust your settings. The overall PG memory footprint to achive the same as SQL-Server on Windows at comparative speed will be quite a bit lower, this is especially true when you run it in headless mode on Linux (no server GUI).
Cost
Yes, quite obviously PostgreSQL wins out here.
Do not forget: It's not just the Server licensing costs we talk about, but also the operating system, and remote access (e.g. additional very expensive Citrix licenses on Windows compared to free SSH access on Linux).
And the other thing: PostgreSQL is OpenSource and Free. If you compare it to SQL-Server, you should not compare it to the Ultimate Enterprise Edition of SQL server (and windows server), but with the version you could actually afford. PostgreSQL however will always be the one free ultimate all-inclusive all-free all-opensource BSD license edition. If you compare the features of PostgreSQL to the Sql-Server Ultimate Enterprise Edition, you'll also have to compare the costs to the ultimate enterprise edition as well (and in 99% of cases, you'll find it not worth the price).
2. Is it ready for primetime?
Yes actually Skype and WhatsApp runs its database on PostgreSQL.
You'll find it has some problems (bad error messages - connection disconnected) when you try to run a 50+ MB insert scripts, but Microsoft's SSMS crashes long before that amount.
Summa summarum, PG probably is more ready for PrimeTime than SQL-Server.
PG has has had paging for almost 10 years now (MySQL as well as SQL-Standard syntax compliant) , SQL server just got that feature with SQL-2012, only standard-compliant.
You can simply xcopy deploy PostgreSQL, works perfectly without admin rights
PG has XML support, just like SQL-Server
PG has JSON support, SQL Server? I believe they are still fully invested in thinking that XML is the future - and purposely crippling the web as a platform. (they are adding JSON support now for SQL-Server 2016 CTP, but it stores data as nvarchar, so you won't be able to set (fulltext) indices on specific JSON-fields like you can on PostgreSQL)
PG has HSTORE support (associative array column), SQL-Server doesn't
PG has array columns. Arrays are a core part of programming, except in SQL Server, where they don't exist (try parametrize an IN-clause)...
PG has had spatial support for a long time, SQL-Server's spatial support is relatively new and relies on .NET extensions and is object-oriented (hello RELATIONAL database? )
PG has the better and faster fulltext search (personal opinion, not backed up by data because it's as clear as it's clear that Google-Chrome is faster than IE 8)
PG has companies offering commercial support, just like Microsoft does for SQL-Server; the advantage is that you actually get your bugs/feature-requests fixed for your money on PG support. You probably know Microsoft support, so I'll spare you any further rants of mine ...
PG has IPv6 support, Microsoft ?
PG performs normal under heavy load, with SQL-Server on Windows we've had some strange issues that appear and disappear at random (Heisenbugs ?).
SQL Server still to this day deploys pessimistic concurrency out of the box, PG doesn't
Compression out of the box. In SQL Server, compression is an "Enterprise Edition+" feature which means you are spending the cost of at least 1 dev in order to get the ability to use compression. Once you have paid for that ability, you still have to figure out how to implement it. Postgres does this for you out of the box, automatically and for free.
Concurrent Index Creation. This yet another feature that SQL Server is capable of doing, but only if you are able to afford Enterprise Edition. Postgres has your back on this and you can leave your wallet at home.
Indexable functions – sargability. In Postgres, you can actually index certain functions and maintain sargability. With SQL Server, BOOM - table scans - adios performance
PG employs Unicode (UTF8) by default. No longer can an incompetent predecessor cripple the entire application with horrible varchar and missing N' for string datatype like they do in SQL Server. The N' syntax still works though, so no compatibility problems if you use them.
PG supports recursive cascaded deletes, SQL server doesn't support that at all (though it actively tells you when you want to put the cascade on the referential constraint)
PG supports GREATEST(value [, ...]), LEAST(value [, ...])
Microsoft: CASE WHEN HugeExp1 > HugeExp2 THEN HugeExp1 ELSE HugeExp2 END) ...
CASE WHEN N > 2 THEN 'ROYALLY SCREWED' ...
or this very maintainable gem
SELECT ( SELECT MIN(Price) FROM ( VALUES (123),(456) ) AS AllPrices(Price) )
PG supports UTF-8 CSV files since ever, SQL server doesn't support UTF-8 CSV files prior to SQL-Server 2014 SP2 (try import the MaxMind GeoIP database - you need to write your own program for SQL-server...).
Drawbacks of PG:
Windows version ain't as stable & fast as the Linux version
No free BI-tool like SQL-Server-Reporting-Services (ok there is Eclipse BIRT, but it runs on JAVA).
On the other hand, for a bit of money you can get stimulsoft reports, which has more export formats than SSRS, and actually renders fine in browsers other than IE as well as IE > 8, unlike SSRS (2005 - 2012 so far, and so far I read SSRS 2014 is no different than 2012).
That said, if your only problem is fulltext-search with SQL-Server Express, you should take a look at Lucene.NET (Apache License).
Also, be careful if you're thinking about using SQL-Express in production:
So far we’ve considered only a few of the most obvious limitations,
but the point is that what seemed like a great idea driven by the
thought of considerable cost savings could end up being an absolute
disaster for your business. Worse, any initial cost savings you gained
from using a free product could be easily negated down the road by the
cost of disaster recovery and the damage to your business’ reputation
while an important database is down. So needless to say, despite the
appeal of its zero dollar price tag, SQL Server Express is definitely
not the right database solution to power your business.
Source: bitwizards.com
This really depends on your DB size and your comfort zone.
SQL Server Express edition is a fully functional free database up to 4GB in size; perfectly adequate for most small to mid-sized apps. (aside - IMHO, MS Licensing was the best thing ever to happen to the open source community!!)
PostgreSQL is a fully functional, free database w/o the 4Gb limitation - however, PostgreSQL follows the Oracle style of SQL and database management. If you're not familiar with that, you can run into some gotchas.
Throwing Dot Net into the mix adds a not insignificant wrinkle. Can you use .Net with PostgreSQL? Yes, but Dot Net will work better with SQL Server. Are you comfortable writing complex ODBC interface code? If yes, PostgreSQL remains a contender. If not, you'll be better off with SQL Server.
From my own personal experience, I find that PostgreSQL runs better on a linux box - lose the windows OS. I'm not very familiar with SQL Server, but I have had no issues with speed or flexibility in working with Postgres. While I can't speak to ADO.NET, I have had much success with using Postgres in the real world (under constant load). In your case, the only variable I could see is integration with ADO.NET. http://www.devart.com/dotconnect/postgresql/ may be a starting point.
As some mentioned above, this really relies on how big your DB will have to be.
Especially the difference price of a CPU-core vs the price of the licence for that core is really heavy.
CPU costs nothing in this comparision.
So if you need 16+ Cores, i would give Postgres a chance.
you will be able to buy a lot more cores for your Postgres server with the saved money.
I am Software developer, and the customers we have, have to pay much more for the Sql Server License than for all the other things together: softwaredevelopment, Clienthardware, Serverhardware, DB-Administration and so on.
Most of the time its cheaper, to buy new hardware, than to optimize your code.
But thats not true for the SQL Server CPU.
(however, you can still invest in Ram/HDDs without raising license costs)
I haven't tried using PostgreSQL with ADO.NET, but from my experience using it with MySQL, you might be better off sticking with SQL Server if you plan to use LINQ to SQL or Entity Framework.
While there are probably third party libraries that will allow you to integrate those technologies with PostreSQL, I think you'll find that things will go a lot smoother with SQL Server. Pretty much all the nice tools built into Visual Studio that make those technologies so easy to work with will only work with SQL Server.
My tables are based on MSSMS 2000 and I'm accessing them with 2008. (I also have VS 2005, 2008, and 2010)
The internal feature does not work with MSSMS 2000.
Please recommend a free third-party/opensource tool, that will only read and not install stuff on the DB server to create a visual display of a subset of the tables (columns, relations etc.).
Before I bit the bullet and bought a full blown license for ERStudio from Embarcadero I used to use the free version of Toad Data Modeler.
It worked quite well, especially considering it's freeware. I haven't used it since Quest took Toad over but Quest has a pretty good reputation in the biz.
SSMS 2008 doesn't have SQL Intellisense when connected to SQL Server 2005! I know of 3 products which enable writing SQL with Intellisense. ApexSQL Edit, Red Gate's SQL Prompt Pro & SQL Assistant. However I am looking for a less expensive solution.
Any ideas?
Embrace Linq and use LinqPad.
Full intellisense using linq on sql servers, fully integrated with c# statements or expressions, for $27.
Those are pretty much the three choices.
If you don't want to pay for an add-in, you could always roll your own (although in the long run it may be cheaper to just pay for an existing solution). Jon Sayce has a good blog post on Building a SSMS Add-In if you want to give it a shot.
My Recommendation: Atlantis SQL Enywhere which is completely free. It works with SQL Server 2005 and 2008. I have been using it for a month now and I am really impressed with its features. Keyboard shortcuts are similar to VS, so makes the transition is really smooth to a new editor.
Some of the features that are worth mentioning.
Intellisense that actually works when using multiple tables and joins with aliases
Suggestion of joins when using multiple tables (reduces time on typing, really neat)
Rich formatting of sql code.
Better representation of SQL plans
Highlights variables declarations while they are used.
Table definition on mouse hover.
Productivity++.
Well, there is a freeware tool that works with SQL Server 2000, 2005 and 2008. Recently we have released dbForge SQL Complete, a free SQL autocomplete add-in for Microsoft SQL Server Management Studio (SSMS).
By the way, on Friday (21 January 2011), we have released a new version of SQL Complete that offers a bunch of new features. All new features are available in shareware Standard Edition (30-days free trial, $49.95).
You also can get all basic functionality, similar to v. 1.0, with Express Edition.
Did the recent purchase of MySQL by Sun and the subsequent buggy releases kill the MySQL brand?
I whole heartedly embraced MySQL when it first came out as I used to be a poor developer and all the RDBMs were too expensive. I have fond feelings for MySQL and their being able to compete with Oracle and SQL Server. I credit the original MySQL team for the existence SQL Server Express. I now use SQL Server Express instead of MySQL for just about everything. First, I do not like Sun and second, SQL Server Express is significantly more robust and 'Enterprise' than MySQL.
The only significant limitations on SQL Server Express are the 4GB db size and the lack of Agent. I find that the size limit is not a concern because by the time the db gets close to that size, the application should either be profitable (and you buy the license), or you should kill the product. The Agent issue is a nice to have, but not critical as you can work around it.
It seems that for db simpletons like me, SQL Server Express is easier to setup and use and is faster and more stable. And for gurus, they will use PostgresSQL...
Resolved:
So basically, we have a bunch of SQL Server fans (albeit open-minded) on stackoverflow. It looks like I'm in the right place. It's sad to see MySQL transform from something you cheer to something you pity. I hope the founders of the original company try a new endeavor. I absolutely do not judge them for selling to SUNW - work is naturally profit driven. I hope we get some people with FU money who don't care about profits who create some products like MySQL used to be. Othewise, MSFT might take away EXPRESS.
Having worked with both, I have to say that the limitations and/or bugs in MySQL were a big turn off for me... I don't like PHP, and while I respect the open source community for their advances with these two technologies I just can't see the elegance in the way either of them have been put together. But don't let my personal taste sway your judgement against MySQL.
I do all of my prototyping in SQL Server Express and most of my clients use full blown SQL Server 2005. The ease of transition from one to the other makes it a no brainer for me - I can take anything I wrote for SQL Server Express and put it straight in without worrying that the syntax might be different. The two limitations don't even really get in the way in a dev environment - it's only when you come to production that they would tend to be a problem.
For me, SQL Server wins the argument hands down.
So far no-one knows what Oracle is going to do to MySQL, not even Oracle.
I've done extensive testing of MySQL and would say that in terms of performance it is about at SQL Server 7.0 level. That is fine if all you need is the performance of SQL Server 7.0
At the enterprise level is simply doesn't compete. If you look at the fanfare surrounding MySQL 5.4 it says "now supports more than 4 processors".
Where MySQL scores highly is that it is so cheap that it makes a scale-out design feasible, in which case the raw power of an individual box simply doesn't matter.
There are quite a few ommissions in MySQL that will trap a SQL Server developer. No CHECK constraints, no index views, no separation of clustered indexes from primary keys. That said, it has a large number of useful features that are very useful to web developers.
The Sun/MySQL guys are remarkably honest about their product. They say exactly what the strengths and weaknesses are. If you are used to some of the big vendors sales tactics this comes as a massive shock. Ultimately this does inspire confidence in using the product because you know exactly what to expect. I would much sooner deal with a product that says "don't do this because this is beyond our limits" than one that says "our product is the cure for cancer" and it turns out it couldn't cure a simple hangover!
Having run both*, I don't think there's much question that SQL Server is overall a more full-featured product than MySQL (although I'd be interested to hear arguments to the contrary).
I wouldn't be overly concerned with MySQL bugs - just run a few releases behind.
Given that you have a Windows server, the only issue I can see is the prospect of acquiring a significant cost if you max out the resources permitted with Express and need to go to a full, licensed SQL Server 2005 (or 2008) instance. If you expect to scale to that level and can't afford the licence, then a free DBMS would seem to be the smart move now, and I'd also look at PostgreSQL, which may be a better alternative.
* and liked them both for different reasons, FWIW
This is a very old article. With MySQL 5.7.20 now, there is no reason to use MS SQL Server or any other commercial licensed database. We switched to MySQL for our high profile financial .Net app six months ago after getting tired with over $50K licensing cost for MS SQL server and this was the best decision ever. No licensing cost and MySQL is highly configurable, small foot print, performance better than MS SQL servers. So far no glitches, no issues. Replication is easy to setup and works like a charm. Did I mention, no money even when running on a 32 core server and having five replication server. MS SQL server costs $3000 per two core (in our setup this would come out to be in six figures). Pure stupidity to throw that money out for nothing. Our own benchmarks, MySQL is performing better than MS SQL server. We really like easily configurable events in MySQL in addition to it being free. Just one another point, it takes only a few minutes to install MySQL while it takes forever to install MS SQL server on a new computer. Don't know what MS SQL server needs to do for hours while MySQL zip file can be unzipped and two line command can configure it within minutes.
MySQL is still going strong, and will continue to.
Note however that SQL Server Express is limited in other ways; specifically, number of CPUs and limitation of RAM usage (1 GB).
Personally I prefer SQL Server over MySQL. It is much easier to work with. And applications properly developed for it can scale to enterprise level if you are careful.
I don't think that you liking Sun or the mySQL 'brand' should influence your choices too much - unless you feel mySQL will be run into the ground by Sun. Look at the performance, management and features.
SQL Express with advanced features looks interesting - now has reporting services. SQL Express also requires a windows license and any associated licensing you might need.
mysql could use http://jasperforge.org/ for reporting.
I like writing tsql more, but for any independent projects, I would likely use mysql due to its cost savings that could be passed on to the client - putting me above the competition. People like saving money, even if it is profit.