RARE user and assistance email list

[Simon Leinen <>]

mc36  writes:
> hi.
> On 8/18/22 13:13, Simon Leinen wrote:
>> Tim Chown writes:
>>> So are there NRENs running SRv6?  I think SURF are running SR, and
>>> GEANT maybe?  But SRv6?
>> We're not using it right now, but our team has been considering it as a
>> possible option for getting rid of MPLS in the core.  We currently use
>> MPLS only to support L2 VPNs, and SRv6 might enable us to run those
>> encapsulated L2 links/tunnels across an agnostic L3 backbone.  (EVPN
>> would be a possible alternative, but we haven't studied that in detail
>> yet.)

> over the obvious fancyness factor and the decreased performance caused
> by the bigger per packet header stack, what is the motivation factor
> to get rid of the mpls dataplane for the srv6?

Note that we really only use the MPLS dataplane for encapsulated L2
traffic, so the performance overhead is not an issue for us (the L2
users are all very low traffic compared to "normal" IP traffic).

The motivation is a general dislike of having connection management
throughout the core, and our lack of deep experience/skills with MPLS.

We mostly see MPLS as a complication that doesn't really buy us
anything, except that we are currently forced to use it for this single
(but very profitable) application - L2 services - which could easily be
implemented at the edges over an agnostic IP core, except our router
vendors used to only implement it efficiently over MPLS.

Another motivation is/was that we don't want to be locked into
MPLS-capable platforms for our core network, even though these days
that's not really a limiting factor anymore (unless you'd want to use
something like Cumulus in the core, but that option is unlikely to ever
become attractive again, unfortunately).

Cheers,
-- 
Simon.