On Tuesday 11 March 2008, Chuck Anderson wrote: > Any my switch has multiple switch fabrics, multiple CPUs, > multiple fan trays, and multiple power supplies :-)
I've found it's easier to justify this on edge and border routers, where the box-level redundancy depends on the connectivity density required and where multiple links would be aggregated into one chassis without the need to multi-home them. Of course, box-level redundancy goes without saying for core routers. However, for core switches in a specific site/PoP, I've always wondered whether having redundant switch fabrics and CPU's was necessary, especially if those switch fabrics and CPU's don't contribute to the forwarding of traffic unless the primary switch fabric has failed. Since all connections to the core switches are mirrored to the corresponding routers, the failure of one switch would be mitigated by the presence of another, with a similar hardware/software configuration. On paper, core switches this redundant make sense. In practice, however, given the "reliability" of today's platforms, having this in a mirrored state seems more like just a nice-to-have. Mark.
signature.asc
Description: This is a digitally signed message part.
_______________________________________________ juniper-nsp mailing list [email protected] https://puck.nether.net/mailman/listinfo/juniper-nsp
