Subject HELP: Misterious IB/FB performance problem on routed TCP/IP network
Author Lukas Zeller

sorry for posting this to ib-support and IBObjects, but I don't know
if it's an IBO or IB/FB problem:

I have a D6 app using IBO 4, which works fine with IB 6 (summer 2000 release)
up to FB 1.0.

However, on a customers network, where the server and the clients are separated
by a few switches and routers, I have a very severe performance problem:

When the app tries to post a dataset, I have a hang of 20 seconds up to 2min.
Then everything continues normally. Immediately (1-10sec) following posts
are quick, but when the app ist left 1 min without activity, next post
again takes 20 secs or more.

Any idea what this could be?

What I tried:
- no change when using another port number (instead of 3050) for IB
- no change between IB 6 and FB 1.0
- no change after moving the IB server to another (heavily RAM equipped)
- TCP-IP network is 100% fine otherwise, many other apps (e.g. MS-SQL)
use the same server and network and have no such problems.
- DNS is fine, name resolves without delay.

For me this feels like some short timeout within IB or IBO that gets
active due to slightly slower response times (ping to the server is
in the 120ms range) because of the routers, and causes a time-expensive
retry of some kind. But I have no idea where to start looking for it...

The exact same setup (same server machine, same client app) works fine
when both are on the same LAN segment.

Any idea that might help is welcome! Thanks in advance!

Lukas Zeller (luz@...)
Synthesis AG, Sustainable Software Concepts & SyncML Solutions