I once worked for a company that was about to go live and they had a managed site with two machines there, plus a number of others at their own location. I identified a problem before they went live when I systematically collected collision rates.
The managed machines had the highest rates of collision. I went over and discovered that because they were physically close, the management team had used a very short 10B2 cable.
Now, everyone knows that you should have cables a multiple of the wavelength or at some suitable distance (about a meter and a bit in this case) apart. Lengthening the cable did wonders for the collision rate.
Moral: Check everything before you go live. The trickle of test code prior to going live would have lulled the bosses into thinking everything was all right. Then when the real users got busy, the retransmit rates would have created a real problem.
Requires Free Membership to View
This was first published in December 2002
Network Management Strategies for the CIO

Join the conversationComment
Share
Comments
Results
Contribute to the conversation