Testing innercity ethernet

I hope this is in the proper forum, and apologize ahead of time if it isn't.

My company recently switched to an innercity ethernet between its two offices which runs over a dual T1 line. Of late, we have been experiencing noticeable speed loss between the offices.

None of us are WAN technicians and are at a loss as to how to begin researching the problem. The company we are dealing with does not seem to be playing straight concerning the speed issue, and we wish to learn more. As we all know, education is power.

We've tried testing the speed across the T1 by using an FTP and observing the results, but we are being told by the installer of the line that this isn't an adequate test. (For the record, the speed results fall very short of the installer's promises.)

I don't even know the proper nomenclature to begin googling efficiently. Any advice would be greatly appreciated.

Hi there,

We us several tests in order to test the speed over links. We do an FTP Download, HTTP File Transfer and also use some software called QCheck which can be used to get a measure of performance of a line www.ixiacom.com/products/performance_app...play.php?skey=qcheck

You do need to take into account several factors when testing the speed. First, if you have been told you have a 1.5Mbps service, remember that this is in Bits and NOT bytes, therefore you need to divide the service of the WAN Connection by 8 to turn it into MB which is what the FTP/HTTP will report. Also, you need to take into account the TCP overhead with the packet headers and TCP Acknowledgements.

Its also best to ensure no other traffic is traversing the WAN during the testing as this can slow down the testing that you are performing.

What equipement are you using on the links ? You can do several Show commands on Cisco Routers to check the interfaces also.

Cheers

Awesome. Thanks, Smurf. I'll toss this into the mix and see what happens.

I'm with smurf on this, we do similar tests here. You can do a straight PC-to-PC file copy under Windows but FTP is better as it introduces less overhead. Try to copy a big file as the error introduced when timing the copy will be proportionally smaller making your result more accurate. However try not to go so big that your source and destination machines saturate their buffers as the throughput will drop off as they start frantically hitting the disk. Thirdly, make sure yours is the only traffic on the link and do it several times and take the average. The calculation is easy, <number of seconds the copy took> / <size of the file in bytes> * 8 = <throughput in bits per second>

One last thing to think about is contention rates = are there other customers using this WAN link, or is it dedicated to yourselves?