Hi all,
I'm currently using a virtual-router instance on an MX80 so that I can
effectively run two different networks with different routing policies
on the same router. I'm already using rib-groups to transfer internal
routes between the two, but am looking to extend that to peering and
select transit routes. Currently, the two policies are as follows:
Network 1 (default instance):
via other routers: transit A full routes, transit B full routes, peering
routes
local: transit C full routes, network 1 internal routes
imported from virtual router: network 2 internal routes
Network 2 (virtual router):
local: transit C static default route only, network 2 internal routes
imported from default instance: network 1 internal routes
I'm thinking of adding provider D with full tables on the virtual
router, so that I have the following:
Network 1 (default instance):
via other routers: transit A full routes, transit B full routes, peering
routes
local: network 1 internal routes
imported: transit C on-net routes only, transit D on-net routes only,
network 2 internal routes
Network 2 (virtual router):
local: transit C full routes, transit D full routes, network 2 internal
routes
imported: peering routes, network 1 internal routes
So the questions I have are as follows:
1) Will I exhaust the FIB, or be close to it now that we're over 512K
routes? Juniper only advertises 1M routes, but from my understanding of
a thread from last year, the 256MB RDLRAM should in theory comfortably
accommodate up to 2M routes along with other requirements
2) Are there any other significant caveats with such a configuration?
3) Would using a bigger, modular MX solve all the drawbacks with FIB
localization?
Thanks in advance for your responses.
-Han
_______________________________________________
juniper-nsp mailing list [email protected]
https://puck.nether.net/mailman/listinfo/juniper-nsp
