The MPLS-OPS Archive[Date Prev][Date Next][Thread Prev][Thread Next] [Date Index][Thread Index][Author Index][Subject Index] Fwd: Urgent Help Needed...
Muzammil, do not forget overhead. Your formula may be correct in general, but every system will have certain overhead (keep-alives, hellos, routing updates). These do not count for customer data throughput but they do count when thinking of overall bandwidth demand. Also, your formula would want to read "number of packets/time". This will give you the average. Some other points to keep in mind: Average throughput may not reflect total possible throughput over a link (think about the effect of TCP windowing and slow start on the average). Also, average over a longer period (say an hour) may be below short peaks. If you are working on throughput, these peaks can be critical depending on the type of traffic. For example, peaks may mean packets get dropped. No real problem for FTP, but definitely a potential problem for streaming media or VoIP. Your hourly average may say that there is plenty of bandwidth available for such traffic, yet you find that VoIP is not working well. All this comes to say that you should beware of being too simplistic in your model. Real life factors that are not shown in the simple formula affect the throughput. I hope this is helpful. Roger Williams Resent-Date: Tue, 18 Mar 2003 08:35:25 -0500------- The MPLS-OPS Mailing List Subscribe/Unsubscribe: http://www.mplsrc.com/mplsops.shtml Archive: http://www.mplsrc.com/mpls-ops_archive.shtml |
|