[c-nsp] Seamless MPLS interacting with flat LDP domains
Robert Raszuk
robert at raszuk.net
Thu May 2 22:15:41 EDT 2019
Radu,
The MPLS in modern DC is none starter purely from technology pov.
In modern DCs compute nodes are your tenant PEs all talking to rest of the
fabric L3. So if you want to roll MPLS you would need to do that to the
compute nodes. That means that with exact match you will see in MSDCs
millions of FECs and millions of underlay routes which you can not
summarize. Plus on top of that an overlay say L3VPNs for the tenant/pods
reachability.
Good luck with operating that scale with MPLS forwarding. Besides while
some vendors of the hosts NICs claim support for MPLS they do that only on
ppt. In real life take very popular NIC vendor and you will find that MPLS
packets do not get round robin queuing to kernel like IPv4 or IPv6 but all
line up to a single buffer.
Only hacking the firmware of the NIC with some other NIC vendor which also
out of the box was far from decent I was able to spread those flows around
so performance of MPLS streams arriving at the compute was acceptable.
Best,
R.
On Thu, May 2, 2019 at 5:35 AM Radu-Adrian FEURDEAN <
cisco-nsp at radu-adrian.feurdean.net> wrote:
> On Wed, May 1, 2019, at 00:15, adamv0025 at netconsultings.com wrote:
>
> > Converting DC to standard MPLS network and all your problems are solved.
>
> Just talk about converting DC to MPLS and you will start having other kind
> of problems.
> IMHO, if DCs didn't massively adopt MPLS is because lack of training and
> what I call "end-user mentality". In lots of cases you are dealing with
> people that when you say "MPLS" they understand "site-to-site L3VPN using
> some magic technology" (that they don't understand). Others do understand
> some tiny bits but see it as a "carrier technology". And there's enough of
> them so that manufacturers take their opinion in consideration. Result =>
> VXLAN.
> Those being said, there are DCs that managed to go the MPLS way, but it
> looks more like an exception rater than the rule. Unfortunately.
>
> --
> R.-A. Feurdean
> _______________________________________________
> cisco-nsp mailing list cisco-nsp at puck.nether.net
> https://puck.nether.net/mailman/listinfo/cisco-nsp
> archive at http://puck.nether.net/pipermail/cisco-nsp/
>
More information about the cisco-nsp
mailing list