On 24/Mar/15 16:17, Adam Vitkovsky wrote:
Fair point I agree but hierarchy is not the way to go. We should not
be forced into troublesome topologies just because FIB space worth of
20k prefixes is considered to be enough for the access layer.
I can't argue with you there; I'm just thinking of all possible options.
I've already told myself Route Leaking is a no-no, but it's still an
option - the one that comes after "Option Z".
Of course, vendors will tell you to run pw's to the nearest PE router
and instantiate a service there, so I'm not turning to them for any
solution other than hardware. I'll have to twist IP into doing what I
want :-).
What we did is we end up stacking boxes into locations that required
more FIB space.
Which is why that 96-port switch from Juniper looks so tasty.
I suppose one could upgrade a 4-port ASR920 to a 12-port, but only if
there are hardware benefits.
Crap, more (good) vendors need to play in this space.
And also multicast distribution is much better with mLDP compared to
full mesh of PW so I don't need to try to sell a hybrid two services
solution for L2VPN+Multicast(via L3VPN).
Loooooove mLDP...
Don't even get me started with SDN :)
To me it's just a woo woo, sorry about the advert but to me it's
exactly like that: https://www.youtube.com/watch?v=sr_EtMhM3fg
Until they have ASICs capable of doing routing/traffic engineering
based on application requests in real time for millions of flows at
multiple 10Gbps rates it's all just a nice dream.
Hehe, woo woo - I'm using that...
Mark.
_______________________________________________
cisco-nsp mailing list cisco-nsp@puck.nether.net
https://puck.nether.net/mailman/listinfo/cisco-nsp
archive at http://puck.nether.net/pipermail/cisco-nsp/
