[Fixed addresses.]
> Remember that SONET is a physical layer protection mechanism so it will
> detect first as against MPLS protection which is L2/L3 (shim). Eventually
> both will. We know that SONET provides restoration to *50ms* but is limited
> to *ring topologies*. It uses extra/more
> resources (which it reserves) than required for protection. SONET's APS will
> try to switch the traffic stream but will not take care of packet
> loss/reordering.
>
> In MPLS, when an LSP is teared down via a link going down, traffic stream is
> to switched to an alternate LSP established on demand (at run time) or a
> preconfigured LSP using a *Faliure Indication Signal (FIS)*. Packet
> reordering is a critical issue because it would degrade (i.e. increase) the
> overall restoration/protection time as detection of out-of-order packets is
> detected at destination and at the transport layer. So you see the
> additional work/time involved. And MPLS layer will take care of packets
> being reordered. (using signaling or other mechanism). Well, this is a very
> interesting issue and lot of people are looking into it since MPLS claims
> "to provide protection/restoration comparable to SONET (50ms)".
Don't know if you examined this specifically in your studies,
but my question would be:
What are the packet loss and re-ordering characteristics for each scheme?
That information helps us understand what might be seen at the higher
layers (including the users' eyeballs/brains) which is what's important,
not some arbitrary number.
Tony
This archive was generated by hypermail 2b29 : Mon Aug 05 2002 - 10:42:36 EDT