Re: [dc] Elevator Pitch

[Thomas Narten <[e-mail]>]

> 1. How many host routes do we reach before we say it "won't scale?" If
> you're talking about leaf nodes in a routing protocol, 100k isn't really
> that big of a deal. Where you do run into problems is between the
> routing and forwarding tables --but that's an area for thought.

Actually, 100K host routes in leaf nodes may well be a Big Deal. It
depends on what type of devices need to understand host routes.

For the scenarios NVO3 is looking at, the edge device includes edge
switches and hypervisors. Yes, such devices can in theory support 100K
host routes (and the routing protocol needed to make that converge).

But at a cost.

And I suspect the cost is too high.

Do you think data center operators will want to run routing protocols
in hypervisors that have to handle 100K host routes? Especially if
such devices will need to be upgraded with better hardware to support
such a load?

Or will they want an alternative approach?

> 2. Why does this mobility need to be at layer 2 specifically? Are we
> assuming DDNS and other sorts of solutions in this space will simply
> never be fast enough/scale far enough/etc?

Like it or not, the key requirement for VM mobility is that the VM's
IP address does not change. That means the VM can't really move from
one IP subnet to another. That means either moving to bigger and
bigger L2s (all under one IP subnet) as the DC expands or the need to
inject /32 host routes.

Neither of those approaches seems particularly scalable/desirable if
you look 10 years down the road and think of 1M+ physical machines in
a DC.

Thomas