RE: Network utilization levels... [7:21884]
In order to accurately gauge a networks utilization you need to establish a baseline and from there evaluate over time, using tools such as VitalSuite, Concord or MTRG to verify your data. Once this is complete, trend analysis can begin and you can begin to formulate an accurate picture of your networks performance. There are, of course, other methods and tools that can be utilized for the purpose of performance engineering. If you are working with a specific application and its affect on overall network performance check out Optimal's Application Expert. HTH, Will -Original Message- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED]]On Behalf Of Chuck Larrieu Sent: Thursday, October 04, 2001 10:34 PM To: [EMAIL PROTECTED] Subject: RE: Network utilization levels... [7:21884] what were your recommendations to alleviate the problem? what was the router model that was being overloaded? would route caching have been of value? Chuck -Original Message- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED]] Sent: Thursday, October 04, 2001 6:27 AM To: [EMAIL PROTECTED] Subject: RE: Network utilization levels... [7:21884] One thing to keep in mind.. A percentage can lie about network performance. If you take a network with a low percentage of traffic based on byte count but a high number of small packets ( Citrix), you can easily have an overloaded router/switch but without the gross load on the wire you might expect. The overload comes from trying to process all the dinky packets which need to be checked, read, forwarded, folded etc. this really does happen as I just did an analysis on a Citrix server farm of over 30 servers hooked to a rather overloaded router. Fun traces showed up on that. MikeS Message Posted at: http://www.groupstudy.com/form/read.php?f=7i=22216t=21884 -- FAQ, list archives, and subscription info: http://www.groupstudy.com/list/cisco.html Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
RE: Network utilization levels... [7:21884]
One thing to keep in mind.. A percentage can lie about network performance. If you take a network with a low percentage of traffic based on byte count but a high number of small packets ( Citrix), you can easily have an overloaded router/switch but without the gross load on the wire you might expect. The overload comes from trying to process all the dinky packets which need to be checked, read, forwarded, folded etc. this really does happen as I just did an analysis on a Citrix server farm of over 30 servers hooked to a rather overloaded router. Fun traces showed up on that. MikeS Message Posted at: http://www.groupstudy.com/form/read.php?f=7i=22035t=21884 -- FAQ, list archives, and subscription info: http://www.groupstudy.com/list/cisco.html Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
RE: Network utilization levels... [7:21884]
what were your recommendations to alleviate the problem? what was the router model that was being overloaded? would route caching have been of value? Chuck -Original Message- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED]] Sent: Thursday, October 04, 2001 6:27 AM To: [EMAIL PROTECTED] Subject: RE: Network utilization levels... [7:21884] One thing to keep in mind.. A percentage can lie about network performance. If you take a network with a low percentage of traffic based on byte count but a high number of small packets ( Citrix), you can easily have an overloaded router/switch but without the gross load on the wire you might expect. The overload comes from trying to process all the dinky packets which need to be checked, read, forwarded, folded etc. this really does happen as I just did an analysis on a Citrix server farm of over 30 servers hooked to a rather overloaded router. Fun traces showed up on that. MikeS Message Posted at: http://www.groupstudy.com/form/read.php?f=7i=22171t=21884 -- FAQ, list archives, and subscription info: http://www.groupstudy.com/list/cisco.html Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
Re: Network utilization levels... [7:21884]
The 70% rule of thumb is in many network design treatises. I know because I put it there. ;-) I put it in the initial Designing Cisco Networks course which grew a life of its own and the rule got listed in all sorts of design books after that. I got it from years of consulting for Network General (makers of the Sniffer). It was a guideline that we used when doing health checks of customer networks. If we saw 70% for a sustained about of time, that was our indication to do some more sniffing to see if the network was perhaps not healthy. A sustained amount of time was generally 10 minutes. The 70% was for WAN, FDDI, Token Ring, and full-duplex Ethernet. The reason for 70% is that if you go beyond that, and a large burst of traffic happens, then it is possible that the burst can't be accommodated and must be queued. You could go beyond 70% if you aren't too concerned about packets getting queued and you don't expect large bursts. The exact number depends on your performance goals and your traffic patterns. On shared Ethernet networks, we got concerned at a lower level, at about 40 or 50 percent, because of collisions on shared networks. Although it completely depends on the number of stations, how often they send, and how large their packets are, at about 40 percent utilization, a typical shared Ethernet network becomes noticeably inefficient and wastes a lot of bandwidth retransmitting frames that collided. This is not a hard-and-fast rule, and I could write pages on why this might not apply on your network, but it's still a reasonable rule of thumb. If you see any documents that present these rules as more than just thumb rules, don't believe them. ;-) Priscilla At 01:46 PM 10/3/01, McMasters, Eric wrote: To all, I'm monitoring a network and I'm seeing high network utilization levels on the main (only) network segment. My question is that I've always heard the rule is 60-70% and your in trouble and can start experiencing network degradation. Now my question is where can I find this in writing? I've been searching CCO and other Internet sites for the better part of a day now and I haven't found anything stating what percentage falls within the best practices / industry standard arena. I know that I have read this info in books, but again I think it just states this as a rule of thumb. I know all networks are different and the saturation number may change depending on how many collisions are occurring etc., but I'm at a loss trying to find this info anywhere. Any assistance would be greatly appreciated!!! Thanks and I hope everyone has a great day!! Eric Priscilla Oppenheimer http://www.priscilla.com Message Posted at: http://www.groupstudy.com/form/read.php?f=7i=21901t=21884 -- FAQ, list archives, and subscription info: http://www.groupstudy.com/list/cisco.html Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
Re: Network utilization levels... [7:21884]
Don't know that you will find any such hard doc but remember link utilizations are percentages over a variable time. Data traffic has wide variations with temporary large peaks or bursts. When you start getting high averages you don't have much room for these bursts and you start dropping packets, hence network degredation. Dave McMasters, Eric wrote: To all, I'm monitoring a network and I'm seeing high network utilization levels on the main (only) network segment. My question is that I've always heard the rule is 60-70% and your in trouble and can start experiencing network degradation. Now my question is where can I find this in writing? I've been searching CCO and other Internet sites for the better part of a day now and I haven't found anything stating what percentage falls within the best practices / industry standard arena. I know that I have read this info in books, but again I think it just states this as a rule of thumb. I know all networks are different and the saturation number may change depending on how many collisions are occurring etc., but I'm at a loss trying to find this info anywhere. Any assistance would be greatly appreciated!!! Thanks and I hope everyone has a great day!! Eric -- David Madland Sr. Network Engineer CCIE# 2016 Qwest Communications Int. Inc. [EMAIL PROTECTED] 612-664-3367 Emotion should reflect reason not guide it Message Posted at: http://www.groupstudy.com/form/read.php?f=7i=21908t=21884 -- FAQ, list archives, and subscription info: http://www.groupstudy.com/list/cisco.html Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]