leonvan
asked on
Domain vs workgroup performance when using UNC named resources
Have a new client. We just installed a workhorse server 2008 (dual quad xeon) and 9 new Windows 7, i7 quad core workstations, and a new gigabit switch. This system should be rocking, but we are not seeing expected performance. The primary software uses UNC resource naming (\\server\path\file).
We have added server/IP to the hosts files and disabled SMB 2.0 (because there is one old XP workstation in the mix), but are still not happy with performance. The expert on the primary software says converting to a domain will make HUGE speed performance improvements.
My question is, can Domain vs workgroup really provide major perfomrnce improvements? What provides the gains (DNS?)?
Thanks,
Leon
We have added server/IP to the hosts files and disabled SMB 2.0 (because there is one old XP workstation in the mix), but are still not happy with performance. The expert on the primary software says converting to a domain will make HUGE speed performance improvements.
My question is, can Domain vs workgroup really provide major perfomrnce improvements? What provides the gains (DNS?)?
Thanks,
Leon
ASKER
The primary application uses a C-Tree database for data storage (with record locking). A large part of the functionality of the application is pointers to image files (in the database). These pointers are UNC format. Could having the local DNS running as part of the domain controller provide significant improvement?
once the workstation knows "server = ip-address" they're done with dns, so, no...
where does it get slow, when you do what?
where does it get slow, when you do what?
ASKER
Basically, any function in the software can be slow (e.g., pulling up a person's record, changing a scheduled appointment, generating a report, printing a report). Since I don't know the inner workings of the software, I can't say what is going on behind the scenes.
that's purely a function of the database itself and not the connectivity between devices, or the processing power of the workstations.
how does the memory on the server look, is it running wild? how about the cpu usage.
if you're not overrunning the resources of the server, it's purely due to bad code on old software - maybe they can migrate you to a real database like SQL?
how does the memory on the server look, is it running wild? how about the cpu usage.
if you're not overrunning the resources of the server, it's purely due to bad code on old software - maybe they can migrate you to a real database like SQL?
ASKER
The server sees no load - 95 plus % idle, memory use is around 50%.
No chance of getting a re-write for at least 2 to 3 years.
No chance of getting a re-write for at least 2 to 3 years.
it seems that your way ahead of the game then, bottlenecked by your software vendors reluctance to improve their product
ASKER
Well, the software vendor is telling us that going to a domain will speed things up dramatically. Running as a domain adds process and overhead, so we are looking to validate the idea before we invest the time and effort into making the change.
Anyone else have any input?
Anyone else have any input?
ASKER CERTIFIED SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
however - with a domain you can use offline files and folder redirection - people take copies of their stuff down to their local machine, and put it back when they're done. that could be a good improvement.
but if everyone is accessing the same file on the server, that won't be able to be used offline or 'checked out' to their workstation.
so it really boils down to HOW are the workstations using the server... are they accessing the same file? or is the application ran by a sql database or something else?
for management, it makes sense to use a domain. for security too. but for performance, it's not going to increase much