I thought I sent this before but we are a multi vendor shop so there is no using EIGRP. That being said we have our core in area 0 only. We have our DR's setup as blocks on our network but the main issue with this is that pretty much everything landed on a single DR so when it reconverges we take outages to user vpn, site to site vpn, user segment, and entire campus infrastructure. Im assuming this is not scaled properly but dont know if each one of those should be attached directly to a core or separate . They are big enough to handle the load but my concern is that they could affect the entire core if they are somehow currently causing entire reconvergence on the DR's they are currently attached to.
On Fri, Aug 14, 2009 at 12:08 PM, Kaj Niemi <[email protected]> wrote: > The problem with making a site an area in ospf is that it limits your > expansion capabilities at that site from day zero - to connect a new site > you must connect it later on to one of the routers (= sites) belonging to > the backbone area instead of perhaps a site that is somehow better (like > closer, cheaper circuits, services are there, etc). That is, unless you > resort to virtual links which imho have no place in a well designed > production network unless you're having to integrate something you > purchased... > > With ISIS you wouldn't have this particular problem as any L2 area will > happily talk with another L2 area. That said I'm certainly not advocating > the OP to deploy ISIS without a bit more understanding what he wants to do > ;-) > > As for priority in a broadcast type of network.. Unless you want to be > really deterministic it most likely doesn't matter who's the DR and who's > not. Sups on the cats won't be having a problem with their DR role unless > you're doing something really silly. Still if it's a /30 or /31 prefix on > your link you pretty much know it's p2p and should configure so, too. > > Assuming the OP wanted L3 all the way to his cat4500s: routed ports (not > SVIs *), /31 addressing on router-router ifs, ifs ospf p2p, area 0 > everywhere, no igp hello/dead tuning - if you cannot rely on media > detection > use bfd with aggressive timers instead, if there is reason to believe that > you're carrying more than a few lans at each site deploy ibgp from day 0 - > you'll save yourself from grief later. > > *) On the 6500s SVIs consume limited vlan space just as routed ports while, > again imho, the syntax for routed ports is easier to understand and one > doesn't have to worry about what vlans are active on what port (assuming > you > don't want to permit 1-4094 everywhere ;)). > > > > Kaj > > > > > From: Joe Astorino <[email protected]> > > Date: Fri, 14 Aug 2009 07:58:20 -0700 > > To: Ryan Hughes <[email protected]> > > Cc: <[email protected]> > > Subject: Re: [OSL | CCIE_RS] OSPF Design Question > > > > A lot of different things come into play here. If this is a simple > topology > > there is no reason we can't have area 0 for everything. If you have a > ton of > > routes going back and forth between the sites, you may make each site an > area > > and leave area 0 for the core and distribution. With that design you > would > > just have type 3 summary LSAs going between the areas. > > > > If you are running point-to-point subinterfaces you will not have a DR > and you > > will not have to deal with priority at all. At the end of the day, it is > more > > of a design decision depending on your environment! > > _______________________________________________ > For more information regarding industry leading CCIE Lab training, please > visit www.ipexpert.com > >
_______________________________________________ For more information regarding industry leading CCIE Lab training, please visit www.ipexpert.com
