On Thu, Dec 29, 2016 at 10:05 AM, Randy Bush <randy@psg.com> wrote:

...

I apparently wasn't very clear. In the layered approach to multiple vendors, you would (obviously) choose your layer definitions to avoid such delicate interdependence.

can you describe in useful detail your operational experience doing this?

I'll certainly try. As one hopefully fairly clear example; at a large (US-nation-wide) metro Ethernet provider, we standardized as follows: L3 devices (aka core, customer edge, and Internet/peering edge routers) were all from Vendor A - These devices spoke OSPF, BGP, and RSVP with each other. L2 devices (aka metro ring switches) were all from Vendor B - These devices spoke STP with each other. L1 devices (aka optical transport) were all from Vendors C or D (individual markets got to choose which, but they could only have one each) - These devices inter-operated with each other at the optical layer. Basic network security was handled by devices from Vendor E - These devices collected netflow data and flagged alerts DNS was handled by software from another vendor on servers from yet another vendor, etc... Is that enough detail to be useful?