PDA

View Full Version : General Routing Protocol Questions


Elshar
08-13-2003, 02:24 PM
Hello,

I'm considering setting up all, or at least most, of my star-os, bsd, and cisco boxes to use some form of dynamic routing protocol.

The issues I'm having though is trying to weigh several factors between RIPv2, and OSPFv2. RIPv1 would be completely useless to me, and BGP is, imo, too intensive for wide-scale deployment.

So, is there anyone that has vast, or at least a deeper understanding between the two protocols? What I'm looking for is the performance/load ratio. Especially since some of our older aps (Which are slowly being phased out) are the old single-board teletronics aps we used to use before star-os came into being.

Right now, my network comprises of 12-16 or so micro-pops, each with at least 2-4 subnets.

Currently, all the routes are statically defined. While I'm just fine with this, I'm starting to run into problems where in order to route a small subnet somewhere I have to manually enter the route every hop along the way every time. And when I want to 'free' that subnet (Due to nat, or unexpected explosion of users on a micro-pop sector), I have to do the reverse.

I've also successfully replicated sections of my network in a sort of lab, and modified it to use various routing protocols. And while it has worked in the lab without issues, its the real-world implementation I'm not sure about.

So, anyone out there have alot of experience with the two, and willing to help me with the pros and cons of the two protocols? I remember someone being fluent in OSPF here. Sort of hoping they'll see this, and help me along. :)

Michael

WISP
08-13-2003, 04:09 PM
OSPF and Zebra don't play well together (wrote the OSPF crash course topic), I learned the hard way.

I moved to a BGP,RIP hybrid.

RIP alone routes based on HOP count instead of metric.

BGP allows us to set metrics.

I would do BGP alone, except that my firewall doesn't do BGP and I choose to dyanamicaly route to it as the last hop.

Elshar
08-13-2003, 04:28 PM
I was originally thinking of just going with RIPv2, where/why do you use BGP?

When I considered BGP, I was thinking of using it between my core routers, and doing RIPv2 along all the ap's, but wasn't sure how viable that'd be, or what sort of affect it would have on the boxes. (all my core routers except one are freebsd, and have pretty big ipf/ipfw rulesets)

Hopefully that made sense.

Michael

WISP
08-14-2003, 10:05 AM
My entire network is firewalled and uses private addressing. I run BGP on all of my APs. I do static routing from the customer to the AP. I do RIP and BGP on my APs in the datacenter to the firewall.

Reason being is that the firewall won't do BGP, otherwise I would do BGP everywhere.

I didn't want to do RIP everywhere because it only routes based upon the shortest path. If one path has more bandwidth than another but has an extra hop, RIP will choose hops over bandwidth. RIP is also limited to 15 hops.

So because of this I use both protocols.

Wayne