[c-nsp] Devil's Advocate - Segment Routing, Why?

[Mark Tinka]

On 19/Jun/20 12:29, Robert Raszuk wrote:
> Saku,
>
> What you are saying is technically true but not realistically important.
>
> Why - the answer is history of PTX.
>
> It was originally designed and architected on the very basis of hardware
> cost and performance when you would only need to switch at rates MPLS.
>
> Well real world showed that you can't sell such box and IP switching has
> been added to data plane there.
>
> Bottom line - I doubt you will find any vendor (from OEM to big ones) which
> can afford to differentiate price wise boxes which would do line rate MPLS
> and any thing less then line rate for IP. And as such IP clearly brings a
> lot of value for simplification of control plane and route aggregation and
> IMHO is a good (well best) universal transport today for all types of
> services from WAN via Campus to DCs (or even MSDCs).

I would agree to some extent.

Even Cisco sort of went down this path with the CRS-3 when they - very
briefly - sold the so-called CRS LSP (Label Switch Processor) forwarding
engine:

   
https://www.insight.com/en_US/shop/product/CRS-LSP/CISCO%20SYSTEMS/CRS-LSP/Cisco-CRS3-Label-Switch-Processor--control-processor/

The goal was a packet processor cheaper than both the MSC and FP data
planes that had been shipping at the time. The LSP card had reduced IP
and QoS scale, as the idea was that all the heavy-lifting would be done
in MPLS.

But, as with the ASR14000 and ME2600X boxes, the LSP card didn't last
long. I guess Cisco figured the FP was a better option, as more and more
traffic started leaving VPN's for the cloud and CDN's.

For us, the PTX1000/10002 make absolute sense, and are options we are
clearly looking at for fixed form factor core platforms as we seek
cheaper 100Gbps ports than we can currently get for the CRS (or even the
MX and ASR9000).

For us, a low-scale IP services box or line card for the core is
something we are very interested in. We don't need rich IP services in
that area of the network, nor do we need rich MPLS services either.
Simple, low-scale IP/MPLS will do.

But for data centre operators who may be anti-MPLS, the vendors will be
stuck between how they develop for them and how they develop for network
operators, for some time to come.

Mark.