6 replies
by cg it
The server isn't powerful enough.
Probably need more memory, SCSI drives instead of SATA, dual processor depends on how many are accessing.
The other factor is the network itself and for that you'll have to do some network testing to find out where the bottlenecks are.
ie one instance of the client running on teh server or a workstation, then there are only a few reasons.
Workstation is underpowered and you are doing a lot of stuff clientside
Network bottleneck and you are passing a lot of data back and forth.
The server has enough spare muscle to cope with a client intensive design.
If it is network you are going to get hammered for query's with large result sets in width and or length.
Might be an idea to install sql tools on the workstation and then run some example queries from the app on both machines.
If you are seeing serious differences that would point to a network problem of some description, if not, then it's the app itself.
NB even if there are substantial performance differences between the workstation and the server this still suggests to me a client intensive design.
One of the classics you often see is people doing things like select * from mybiggesttable and then using ado's locate (which works clientside !) I recently ported a desktop backend which was properly optimised to work like this. In client server though it was horrible. 3.5 seconds for the locate, 0.1 when I shifted to select * from mybiggesttable where ID = ???
HTHs
What network protocols are you using? When you connect to a local SQL machine you are almost certainly using Shared Memory(which is very fast, but can only be used locally). When you connect through the network you are likely using TCP/IP or named pipes, so check your protocols. Likely though, it is the network. What are the specs on the server you are using?
Tim
by uberg33k50
by support
This discussion is closed to new replies.
Trending Networks posts
Sort by Trending