----- Original Message ----- From: "Chuck Larrieu" To: Sent: Monday, April 30, 2001 1:55 PM Subject: WAN Congestion - Cloud technology - Theory vs Reality [7:2618] > dropped/refused entry because of congestion on the cloud. This in turn leads > to the retransmission of dropped packets/cells, which in turn leads to more > congestion, in a never ending spiral ( in theory, at least ) > > Reality: This gets into sizing of WAN links / CIR's / CBR's > > I am a bit curious. Anyone here have any real world experience with this > kind of thing happening? I can see how this can happen in theory. In > reality, carrier cloud congestion is not such that it would likely lead to > this kind of result, is it? > Real life experience shows that frame relay circuits with 70% or more port utilization create an environment for retransmission. However, based on packet capture and analysis, I see the retransmissions as much as I see connection timeouts because the latency of the circuit increases when the load maxes out, so it's not the "never-ending spiral" as one might expect. I haven't seen any production ATM networks drop cells. In reality, regarding carrier cloud congestion, domestically the carriers claim they overbuild the capability of their service backbone precisely to prevent such congestion problems and I'd tend to believe them because of my experiences with outages that never affected my circuits. Trans-oceanically (is that a word?), there are some serious contention problems especially from South America and Asia/Pac to the US. > So if the above premise is something that can and does happen regularly, > what does the carrier do - just massive dropping of packets / cells until > the problem disappears, probably after hours that day? > If it is because of a failure on the part of their network, they try to repair and re-route traffic to eliminate the data drops. SLA's usually cost the carrier money at a certain point in the failure window, and a reputation for bad reliability is a difficult issue to overcome in this marketplace. If it is because the customer has overutilized links, they try to sell the customer more bandwidth. > Any experience? > > Chuck > > One IOS to forward them all. > One IOS to find them. > One IOS to summarize them all > And in the routing table bind them. > > -JRR Chambers- > FAQ, list archives, and subscription info: http://www.groupstudy.com/list/cisco.html > Report misconduct and Nondisclosure violations to [EMAIL PROTECTED] Message Posted at: http://www.groupstudy.com/form/read.php?f=7&i=2634&t=2634 -------------------------------------------------- FAQ, list archives, and subscription info: http://www.groupstudy.com/list/cisco.html Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
