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 |
AskSQLTeam
Ask SQLTeam Question
0 Posts |
Posted - 2000-11-22 : 20:08:33
|
Richard writes "Which is the way to better performance with Microsoft SQL Server 7.0: using a single database with lots and lots of tables or logically grouping data into multiple databases each with a significantly fewer number of tables? The system is comprised of one server with two hard drives that are mirrored. With multiple databases come multiple log files. Does having multiple transaction log files impair a recovery if needed? Also, is there any benefit in using file groups with this configuration?" Article Link. |
|
imei
Starting Member
3 Posts |
Posted - 2007-04-11 : 17:33:46
|
It works well for OLTP applications to break down one single database into small databases to preserve the data and maintain the on-going transactions. I'm not sure the benefits would be significant for OLAP applications and the extra work required in the maintenance and integration would be worth of the trade-off.Especially in the situation when the size of the database is considered to be small (in this case) and the data is mostly RO.Any comments?Thanks,IChen |
|
|
imei
Starting Member
3 Posts |
Posted - 2007-04-11 : 17:35:23
|
quote: Originally posted by Anonymous Fault Tolerance <P>Suppose you are expecting 10,000 clients. 1) Would you still recommend a database instance per client?2) Suppose you have a couple of client database instances spread across a couple of servers. I am not very familiar with clustering capabilities in Win2K. Is providing a fault-tolerant solution still possible without much effort?I think with a single database instance implementation, servers could be clustered in an active-to-active configuration providing a fault-tolerant and load-balanced solution in the ASP industry. When creating client database instances and spreading them across a couple of servers, providing a fault-tolerant solution may get more difficult because each server does not look identical. Is this true?
|
|
|
|
|
|