Please start any new threads on our new
site at https://forums.sqlteam.com. We've got lots of great SQL Server
experts to answer whatever question you can come up with.
| Author |
Topic |
|
Dinky
Starting Member
37 Posts |
 Posted - 2009-08-17 : 23:05:51
|
| We are creating a new enterprise level software (web application) to be installed on 2 node 2005 SQL Server SAN cluster. The DB is going to be running live OLTP application and live reports. The DB is expected to be about 150 GB to start with. XML and image (< 1 MB) files make 2/3 of database size. DB is expected to have max 50-100 concurrent users. DB has data auditing running - all transactions being written to history tables (can grow to terabyte in 2 years time).Are these really highly desired install features? - Partitioning, More CPUs supported, Fast Recovery, Indexed View Matching, Parallel processing of indexing operations, Online Restore, Advanced SSIS features, Additional tuning options for the accuracy, performance, and scalability, Text Mining for analysis via reporting and OLAP.Trying to determine weather we really need SQL Server Enterprise or Standard edition is good enough for our requirements. Currently in UAT application is running reasonably well.Even if we don't need Enterprise now, are we going to need it in near future for performance, scalability, overall maintenance of db in long run |
|
|
russell
Pyro-ma-ni-yak
5072 Posts |
Posted - 2009-08-18 : 00:37:34
|
| The decision will probably come down to # of CPUs. Have you stress tested? Partitioning will be a nice feature if/when tables get very large |
 |
|
|
Dinky
Starting Member
37 Posts |
Posted - 2009-08-18 : 09:32:03
|
| Which features out of these are generally used and for exactly/practically what purpose:Partitioning, More CPUs supported, Fast Recovery, Indexed View Matching, Parallel processing of indexing operations, Online Restore, Advanced SSIS features, Additional tuning options for the accuracy, performance, and scalability, Text Mining for analysis via reporting and OLAP.Can we say, Standard edition's features are generally enough in 95% cases. |
|
|
|
russell
Pyro-ma-ni-yak
5072 Posts |
Posted - 2009-08-18 : 10:39:07
|
| I don't know about 95% but if you can get by with 4 CPUs then you're probably fine with SE.If you're planning on growing a single DB to 1 TB then I think EE is the way you're going to want to go. |
|
|
|
Dinky
Starting Member
37 Posts |
Posted - 2009-08-18 : 11:28:54
|
| Thanks Russel. This is helping me a lot. Please elaborate a little more.I think 2 more important and more commonly used ones are:Partitioning, More than 4 CPUs - What do you think about these EE features, are we going to need these 2 items.Not sure on these.Fast Recovery, Indexed View Matching, Parallel processing of indexing operations, Online Restore, Advanced SSIS features, Additional tuning options for the accuracy, performance, and scalability, Text Mining for analysis via reporting and OLAP.Please advice from your experience. Can we atleast say, we should be good on SE to start with may be for a year or so and then move to EE when really needed. |
|
|
|
russell
Pyro-ma-ni-yak
5072 Posts |
Posted - 2009-08-18 : 13:20:00
|
| yes, but be prepared early if migrating to EE becomes necessary.for me, we have several servers with 32 cores. obviously we need EE on those. they are very high volume OLTP servers and one multi-TB data warehouse. partitioning has made some issues much more manageable. we have a few tables with several hundred million records in them. at the time we upgraded our hardware, however, EE was an easy call because we knew we needed the CPU power.for our other apps, we use SE as we only need 4 or fewer CPUs.the best way to predict what you will need is to measure what you use now. for a new app, this is impossible, but you CAN stress test it, by simultaing a production level load and see where you stand. with regard to CPU in particular, but also RAM and I/O.you probably want a good plan for archiving those history tables. they will grow to have millions of records i assume. these are perfect candidates for partitioning. however, that is something u can do on down the line. just remember that the more they grow, the more time it will take later to move the data if you choose to.hope that helps a little |
|
|
|
russell
Pyro-ma-ni-yak
5072 Posts |
Posted - 2009-08-18 : 13:23:41
|
| also...maximize the RAM. with SE you still can utilize all the RAM the OS supports. Lots of RAM and proper indexing goes a long way towards preventing CPU bottlenecks. |
|
|
|
|
|
|
|
|
SQL Server 2005 Enterprise v/s Standard