For me, it was a bit of an experiment, but I have ended up liking
it. Yes, it does add some overhead, but I didn't have to add routers
to be the route reflectors - I just chose two routers which provided
good geographic redundancy balanced with being as well-connected as
possible to the rest of the routers and checked the "route reflect
to peers" box. Route reflecting is really no more intensive than
just BGP peering; probably most already know this, but the only
different between a route reflector and a non-route reflector is
that at route reflector is allowed to break the iBGP rule of not
disseminating routes learned from one peer to another peer. One of the things I really like about using BGP for access prefixes is that I don't have to mess with filters or using non-backbone areas and area-ranges to summarize pools used for things like PPPoE. It's nice that more recent versions of MikroTik automate adding the U route of a summarized area-range after the first connected route shows up, but with BGP, I simply add the prefix to Networks and it's done. Another advantage, albeit a "band-aid" one is that if I'm having some link quality issue that is ultimately causing OSPF to lose adjacency (packet loss causing dropped Hello's, for example, or some jackass carrier providing a circuit that upgrades their platform and they don't read the release notes and multicast gets dropped...), I can deploy a small handful of static routes to improve stability slightly until I can resolve the issue (just a small time saver). Obviously, none of this functionality REQUIRES the use of BGP and it can all be done using OSPF. Indeed, while I'm using OSPF + iBGP in my WISP, the telco I'm also the network architect/engineer at uses only OSPF as the IGP and we have thousands of internal OSPF routes and dozens of routers in the backbone area (along with others in non-backbone areas) and it's extremely stable. I think its easy to misinterpret problems which manifest themselves as OSPF issues, but are really just OSPF reacting to some other condition; the canary in the coal mine, if you will. <rant> If you're having issues with OSPF losing adjacencies or changing from full to down or full to init, you've got some problem with the link. Period. OSPF is not the problem. OSPF has been stable in MikroTiks since 3.x.</rant> Jesse DuPont Network
Architect Celerity
Broadband LLC Like us! facebook.com/celeritybroadband On 8/26/16 1:16 PM, Faisal Imtiaz
wrote:
|
- Re: [AFMUG] (OSPF + ibgp) / formerly Mikrot... Faisal Imtiaz
- Re: [AFMUG] (OSPF + ibgp) / formerly Mikrot... Mike Hammett
- Re: [AFMUG] (OSPF + ibgp) / formerly Mikrot... That One Guy /sarcasm
- Re: [AFMUG] (OSPF + ibgp) / formerly Mikrot... Josh Reynolds
- Re: [AFMUG] (OSPF + ibgp) / formerly Mikrot... That One Guy /sarcasm
- Re: [AFMUG] (OSPF + ibgp) / formerly Mikrot... Josh Reynolds
- Re: [AFMUG] (OSPF + ibgp) / formerly Mikrot... David Milholen
- Re: [AFMUG] (OSPF + ibgp) / formerly Mikrot... Paul Stewart
- Re: [AFMUG] (OSPF + ibgp) / formerly Mikrot... Mike Hammett
- Re: [AFMUG] (OSPF + ibgp) / formerly Mikrot... Paul Stewart
- Re: [AFMUG] (OSPF + ibgp) / formerly Mikrotik O... Jesse DuPont
- Re: [AFMUG] (OSPF + ibgp) / formerly Mikrotik O... Faisal Imtiaz
- Re: [AFMUG] (OSPF + ibgp) / formerly Mikrotik O... Mike Hammett
- Re: [AFMUG] Mikrotik OSPF weirdness Paul Stewart
- Re: [AFMUG] Mikrotik OSPF weirdness Ken Hohhof
- Re: [AFMUG] Mikrotik OSPF weirdness Dennis Burgess
- Re: [AFMUG] (OSPF + ibgp) / formerly Mikrotik OSPF w... Faisal Imtiaz
- Re: [AFMUG] (OSPF + ibgp) / formerly Mikrotik O... Bruce Robertson
- Re: [AFMUG] (OSPF + ibgp) / formerly Mikrotik O... Faisal Imtiaz
- Re: [AFMUG] (OSPF + ibgp) / formerly Mikrotik O... That One Guy /sarcasm
- Re: [AFMUG] (OSPF + ibgp) / formerly Mikrotik O... Bruce Robertson