On 17/Apr/16 19:34, Saku Ytti wrote:
> There is upside to that strategy, if you drop from long list all > devices which have a problem, but problem you can workaround with, you > don't have to build a network. As with any network build, you have to make compromises. There are many boxes I have not bought because they can't do something we consider basic. Luckily, I've not been in a situation (yet) where I've not had an alternative. But if that day would come, I'd make the hard choice. > That appears to support my argument. You have flat pool of prefixes, > the more VRF you divide it with, the less prefixes per VRF? In the 2000's, l3vpn's were more popular than l2vpn's. For networks that sold a ton of l3vpn's, you'd have had to consider how many l3vpn's you want to sell vs. how many Internet routes you want in an Internet VRF. Assuming some platforms still have such a restriction, it's too much admin. for me. But again, that's just me, since several others have made this work for them. > That is my belief, yes. But I've never tested past few thousand BGP > sessions. Usually there isn't any hard limit, it's just config gets > slow to parse, boot time becomes problematic etc. Fair enough. Mark. _______________________________________________ juniper-nsp mailing list juniper-nsp@puck.nether.net https://puck.nether.net/mailman/listinfo/juniper-nsp
