SRv6 Capable NOS and Devices

Mark Tinka mark at tinka.africa
Mon Jan 17 08:13:56 UTC 2022



On 1/17/22 09:57, Brandon Butterworth wrote:

> Isn't the argument here that if it's in most chip sets already it might
> reasonably be expected to be a standard low end feature by now, along
> with IPv6?
>
> That it isn't may be why people are open to SRv6 (I'm assuming some are
> based on this discussion) - if they have to pay extra they only want to
> do so where they are generating revenue from it, the end points.
>
> Complexity and architectural cleanliness are not a consideration, if a
> vendor makes a box that does the job at the right price there is a high
> risk people will buy it.

There are other things that are required to support MPLS services than 
just chip and software capability. Control plane, buffer memory, queue 
depth, e.t.c.

That is why you would not find a lack of MPLS support in Ethernet 
switches... you would find "broken" support, because the rest of the 
hardware is not designed to deliver the same level of MPLS service scale 
that a high-end router would; and the vendors make that very clear. An 
Ethernet switch running MPLS can probably be an excellent P router, but 
won't be an awesome PE router.

There are still some IP routing features we consider "basic" in bigger 
routers that attract extra costs in Ethernet switches. Never mind MPLS.

Let's not confuse "MPLS no longer being expensive" with "MPLS being a 
low-end feature".

I find it hard to believe that SRv6 support will be enabled on low-end 
devices like Ethernet switches by the traditional vendors. But hey, I 
have been wrong before.

Mark.


More information about the NANOG mailing list