General discussion

Locked

Network Performance

By rustylh ·
FED UP!
Background: Accelar 1200 (Switch Mode) feeds about 10 Baystack 450 switches via fiber links. WIN NT domain, 3com nics, DHCP, autonegotiation on all devices.
We are seeing very poor network performance. Very high cpu utilization on Accelarand sometimes have to reboot it. Some workstations experience NT Drive disconnects during file saves, slow access to server. Using DHCP and TCP/IP, some ipx/netbui may be still floating around. Any idea's?

This conversation is currently closed to new comments.

3 total posts (Page 1 of 1)  
Thread display: Collapse - | Expand +

All Comments

Collapse -

Network Performance

by McKayTech In reply to Network Performance

Whenever I see the phrases "autonegotiation" and "poor network performance" together, I usually assume they are related and the first thing I do is hard-code full-speed, full-duplex on all non-workstation interfaces.

That having been said, a "jabbering" NIC could cause this sort of problem but there are a bunch of other possible causes as well and the only way to really know what is going on is to capture a stream of packets and analyze it.

In one system I recently analyzed, the problemwas simply insufficient RAM on the core switch. Because there was not enough RAM to create buffers for all the inbound packets, about 25% were getting dropped and re-sent which started a vicious cycle of increasing traffic. A re-boot was a temporary fix until the buffers filled up again. But we found it first by analyzing a packet capture and seeing lots of re-sends.

paul

Collapse -

Network Performance

by rustylh In reply to Network Performance

Poster rated this answer

Collapse -

Network Performance

by rustylh In reply to Network Performance

This question was closed by the author

Back to Networks Forum
3 total posts (Page 1 of 1)  

Related Discussions

Related Forums