[c-nsp] spanning-tree for local switching on ASR920

[James Bensley]

On 19 October 2017 at 09:38, Mark Tinka <mark.tinka at seacom.mu> wrote:
>
>
> On 19/Oct/17 10:24, Gert Doering wrote:
>
> So, how do you bridge together two ports on an ASR1k, with STP? ;-)
>
> I do understand the bits about no global VLAN significance, etc.,
> and tieing bridge-groups to pseudowires, etc. - I just want the more
> basic stuff to be more explosion-robust when the customer plugs in
> things wrongly.  OTOH, the box is doing well, circulating 330.000 PPS
> in that STP loop, with barely any CPU load [these are IP/ARP broadcasts
> so the CPU is at risk of being told about them]...
>
>
> Good question, couldn't possibly tell you that.
>
> We typically don't support such topologies at Layer 2 in our network due to
> the very problem you describe. We'd run both ports independently for the
> same customer, as for such use-cases, they are either looking for redundancy
> or additional bandwidth; or both.


Sometimes beggars can't be choosers but this basically ^

We wouldn't offer dual connections to the same layer 3 edge device as
a "resilient" service nor have it participate in layer 2 service if it
is layer 3 edge. I'd stick a switch in place, the FW could have two
links to the switch and the switch can participate in STP and have one
uplink to the ASR920/PE for layer 3 termination/upstream.

Cheers,
James.