Discussion Locked

Intel Switch

0 helpful
9 replies
Closed to new repliesThis discussion is closed. It stays readable, but it takes no new replies.

We are using Intelexpress 510T Switch with stack & matrix interconnection. We hv 70 nodes (40 Dsktop & 30 thin client) 5 servers (2 NT 4 & 2 W2k Adv Server 1 Novell.). We hv forced all to 100mb & full duplex. Still our network is slow & also we get time out when pinging. Our network is in ECAT5 cable & is tested with Fluke meter. Also we checked all parameters & hints given by intel in their website. Can anyone give us some clues on the above.

Thanks

9 replies

by mpdcsup

What protocols are you running? What's the topology of the network? What's the load? The utilization? Why do you find it necessary to force the NICs to 100/Full? Is your broadcast traffic eating you alive?

by jackieb

Sounds like you need to have someone well practiced on an analyzer come take a look at your network to determine where the bottleneck is. From what you have stated, it does not seem like a bad set up but obviously you are experiencing problems. There are a lot of things that could be eating bandwidth, too many to speculate on here, but an experienced person wielding a good network analyzer could probably identify your trouble spots fairly quickly, and propose a fix. One note, you should try to find someone experienced on the equipment you are using so they will be comfortable with it. One thing I just thought of, yoou stated stack and matrix interconnection, whjile I am not sure what the 'm,atrix interconnection' is, it soounds like another way of inter connecting different switches. If you have this interconnection and the stack cables installed, they could be causing a loop situation, causing a lot more traffic than normal. Check to see if you have redundant paths between the switchsegments and if you do either remove them or make sure that 'spanning tree' is enabled on all yoour switches.(Note that each 'blade' may be considered a separate switch for configuration purposes) Good Luck. JackieB

by lance.pederson

I know it says your problem is already answered but I have worked with those particular Intel switches extensively. I just have to get my 2 cents in. Have you ensured that the firmware is the same on all the switches? I made the mistake once of adding an additional switch to a stack once without updating the firmware and I ran into problems like you are talking about. Hope you got it fixed.
Lance Pederson
Convergence Northwest

by fred_e944

Are all your computers using the same NIC? We installed a switch and had lots of problems until we forced all the NICs to half-duplex. Full-duplex can cause problems when a mix of brands of NICs are coexisting on the network. Check that the cables are not located near sources of electrical or magnetic interference. I don't know what sort of meter you used, but many simple test continuity with DC and do not truly guarantee operation at data speeds.

The timeouts on pinging can happen if a card or cable is marginal at 100 mbps. Try slowing to 10 as a test. If response is quicker or the same you are having dropped packets. It takes a lot longer for a request to time out and be resent than to send it successfully at lower speed.

This discussion is closed to new replies.