|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
||||||||||||||||
|
|||||||||||||||||
|
|||||||||||||||||
|
|||||||||||||||||
|
|||||||||||||||||
|
|
||||||||
|
|||||||||
|
|
|
|
Company
|
|
 SQL LINKS DEVELOPER SUPPORT
SEARCH Borland Online/Developer Support
There seem to be performance differences between IPX and TCP against MS-SQL. =============== Thanks for your response. It prompted to perform some additional tests, which yielded some strange results. With TCP/IP as the Win95 client network layer (set in Control Panel/Network/Configure->Protocol->Properties/Bindings->Client Service for Microsoft Networks), I could select any of four Default Net Libraries in the MS SQL Client Configuration Utility -- Named Pipes, NWLink, MultiProtocol and TCP-IP -- and get fast results. (IPX/SPX was also installed as a protocol but it was not bound to Client Service for Microsoft Networks; it was being used for a concurrent connection to a different NetWare server). When I set my Win95 network layer to IPX/SPX w/NetBIOS enabled, however, the only consistent result was with TCP/IP as the SQL Server net library -- the application couldn't connect. Named Pipes and Multi-Protocol were sometimes fast (27 seconds for a big job) and sometimes slow (240 seconds for the same job). The results were consistent as long as I did not change the configuration, but making changes and rebooting would yield inconsistent results. I'd be very grateful if you or anyone else could help me understand what's causing these discrepencies, or point me towards a good resource, or at least tell me what SHOULD be happening. I've checked MSDN, SQL Server Books Online and TechNet, but have not been able to get to the bottom of this. =============== >>I could select any of four Default Net Libraries in the MS SQL Client Configuration Utility -- Named Pipes, NWLink, MultiProtocol and TCP-IP -- and get fast results.<< By fast results, do you mean general workstation performance, or database performance? >>(IPX/SPX was also installed as a protocol but it was not bound to Client Service for Microsoft Networks; it was being used for a concurrent connection to a different NetWare server).<< I do this also, binding IPX/SPX>adapter>Client for Netware and TCP/IP>adapter>Client for Microsoft. The MS client is the primary login. Overall performance is predictable and consistent. >>When I set my Win95 network layer to IPX/SPX w/NetBIOS enabled, however, the only consistent result was with TCP/IP as the SQL Server net library -- the application couldn't connect.<< This makes sense, because the DB client using TCP/IP needs Win95 to have TCP/IP running. >>inconsistent results<< >>what's causing these discrepencies<< There are a lot of factors involved. What else is the server doing? How many users on the network? What about network traffic? Is there a cabling problem? (That one could drive you nuts!) What else is installed on your PC? Is it a "clean" Win95 install, or has a lot of software been added/removed over time? Do the machines have enough memory? And so on... As far as the inconsistencies you mention, one way to get to the bottom of it in any reasonable amount of time would be to remove as many variables as possible -- set up a "fresh" server and workstation as the only computers on the network and repeat your tests. But that's still a lot of work. In the long run the best thing to do might be to just bite the TCP/IP bullet. MD MSSQL TECH PUB
|
|
||
|
|
|
|||
|
|||||
|
|
||||
 |
|||||
|
|||||
| Made in Borland® Copyright© 1994-2003 Borland Software Corporation. All rights reserved. Report Piracy, Legal Notices, Privacy Policy Last Modified Monday, 26-Mar-2001 16:27:08 PST |
|||||
