On Sat, Sep 23, 2006 at 08:59:29PM -0500, Eric Helm wrote: > > Richard A Steenbergen wrote: > > > > The Foundry MPLS support is "extremely > > simplistic" compared to other routers, but functional. > > Our company is about ready to purchase 17 of the MLX series and begin > moving towards an MPLS backbone. Unfortunately, due to poor budgeting by > management, we have really no options to look at other vendors. We have > very little MPLS experience with any vendor, so I'm not sure where to > begin comparing. Do you have some examples of "simplistic" MPLS support > in comparison to vendor J or C?
By simplistic I mean, it supports all the basic functionality to actually be considered real MPLS support (P/PE roles), but it doesn't do a lot of the advanced features you would find on a carrier grade core router. You can turn on label switching, configure basic LDP/RSVP signaling, configure an LSP (static or signalled), configure VLL/VPLS or forward IP over an LSP, and do basic TE (cspf, preferred paths, tie breaking, etc). Things it doesn't currently do... autotunnel, any auto fast reroute, FRR bypass/facility protection, fate sharing, soft preemption, graceful restart, ISIS TE, P2MP, BFD, configurable autobandwidth, etc (non-exhaustive list, just my first pass looking through it). Some of those are features that any self respecting carrier who seriously uses MPLS would really require, but Foundry in the core was never really an option for them anyways. :) The positive side of that is, the Foundry MPLS implementation is simple and efficient for people who do need some MPLS support (say for example an edge box doing VLL/VPLS customer termination, where real carriers can find serious value in these boxes), or for people who don't need the advanced features in their network. That said, if you really don't understand MPLS and understand WHY you need MPLS, I would urge you not to deploy it. It doesn't do magic, it is a tool for managing complex networks and thus it is complex itself. The standard test for the "should I be using this" question applies here, if you even have to ask the answer is probably no. :) -- Richard A Steenbergen <ras at e-gerbil.net> http://www.e-gerbil.net/ras GPG Key ID: 0xF8B12CBC (7535 7F59 8204 ED1F CC1C 53AF 4C41 5ECA F8B1 2CBC)
