Hi Yakov, On 11 Feb 2014, at 14:21, Yakov Rekhter <[email protected]> wrote:
> Giles, > >> Hi Thomas, >> >> There are two key differences between E-VPN and virtual-subnet wrt scaling: >> >> 1) E-VPN supports interconnection of routers. virtual-subnet >> doesn't. One of the key use-cases for a multi-point L2 service >> is to interconnect routers. > > From draft-ietf-l2vpn-evpn-04: > An EVPN instance comprises > CEs that are connected to PEs that form the edge of the MPLS > infrastructure. A CE may be a host, a router or a switch. The PEs > provide virtual Layer 2 bridged connectivity between the CEs. > > From the above quote it should be clear that while it is true that > "E-VPN supports interconnection of routers", it is also true that > E-VPN supports interconnection of hosts and/or switches. Indeed. And I wasn't claiming that it didn't. What I was trying to make clear was that E-VPN provides two ways to scale that are missing in virtual-subnet: 1) you can use it to interconnect routers. So you get one FIB entry per router. 2) you can use it to interconnect PBB switches. So you get one FIB entry per switch. AFAIK virtual-subnet only provides the model where you interconnect hosts and get one FIB entry per host. Giles > Yakov. > >> >> 2) As well as E-VPN we are standardising PBB E-VPN in L2VPN. PBB >> E-VPN only advertises provider MAC addresses in BGP. Client MACs >> are learned/cached in th e data plane. >> >> Giles >> >> On 11 Feb 2014, at 09:56, Thomas Morin <[email protected]> wrote: >> >>> Hi Giles, Thomas, >>> >>> I would echo Robert's questions below, and ask an additional question: adve > rtising non aggregated host routes (v4 /32 routes or v6 /128 routes) in BGP > VPN > v4 routes is not fundamentally different from a scaling standpoint than > adverti > sing MAC addresses in E-VPN BGP routes. >>> What would be the reason to believe it would be an issue in one case but no > t in the other ? >>> >>> -Thomas >>> >>> 2014-02-11, Robert Raszuk: >>>> <changing subject to reflect more broader l3vpn related topic> >>>> >>>> Hi, >>>> >>>> Could those who claim that that sending /32 or /64 or /128 in BGP mainly w > ithin contained DC zone environment will not scale be a bit more precise and > ki > ndly indicate what the real problem is ? >>>> >>>> * Which control or data plane element will not scale ? >>>> >>>> or >>>> >>>> * Which part of BGP state machine will not scale ? >>>> >>>> Just curious .... >>>> >>>> Cheers, >>>> R. >>>> >>>> >>>>> >>>> On Tue, Feb 11, 2014 at 3:03 AM, Thomas Nadeau >>>>> >>>> <[email protected]> wrote: >>>> >>>> Thomas, >>>> >>>> I too object to it's adoption based on the /32 point Giles made. >>>> >>>> >>>> On Feb 10, 2014:2:29 PM, at 2:29 PM, Giles Heron <[email protected]> w > rote: >>>> >>>>> I don't support adoption of this draft as a WG item (speaking as a non-a > uthor but name-checked commenter). >>>>> >>>>> The draft has a major limitation (no support for interconnecting routers > , but only for interconnecting hosts), and I'm unconvinced that passing /32 > hos > t routes around in BGP will scale. >>>> >>>> >>> >> >> >> >
