On Wed, 2004-02-25 at 13:34, David Meyer wrote: > Is it that sharing fate in the switching fabric (as > opposed to say, in the transport fabric, or even > conduit) reduces the resiliency of a given service (in > this case FR/ATM/TDM), and as such poses the "danger" > you describe?
Our vendors will tell us that the IP routing fabrics of today are indeed quite reliable and resistant to failure, and they may be right when it comes to hardware MTBF. However, the IP network relies a great deal more on shared/inter-domain, real-time configuration (BGP) than do any traditional telecommunications networks utilizing the tried and true technologies referenced above. Yesterday we witnessed a large scale failure that has yet to be attributed to configuration, software, or hardware; however one need look no further than the 168.0.0.0/6 thread, or the GBLX customer who leaked several tens of thousands of their peers' routes to GBLX shortly before the Level(3) event, to show that configuration-induced failures in the Internet reach much further than in traditional TDM or single vendor PVC networks. The single point of failure we all share is our reliance on a correct BGP table, populated by our peers and transit providers; and kept free of errors by those same operators. -- JSW
