NTAC Peering Working Group

[Nick Buraglio <>]

I've seen this quite a lot in the wild. There are a lot of smaller ISPs with older devices out there like 6500's with sup720's that smaller regional providers are still using - even with dual stacked networks. They take fabric peerings, have any number of full transit providers and then either prune the routes via AS-Path or take default plus direct from the transits. The issue of capacity is solved since they can fail over or lose their IXP bi-laterals or any of the transits and always fall back to default. 

Even with the newer devices they tend to operate on the cheap just to stay afloat and since their routing model works well they typically optimize for throughput and port count rather than route table size. MTB replacement is significantly longer in this space. As with some smaller universities, their budget goes toward the most visible things that can be easily augmented such as raw transit.  Given the choice of adding more capacity because your upstreams run too hot or forklift replacing the equipment to take more routes, capacity wins every time up until the interface speed or port count doesn't meet needs. 

nb

ᐧ

---
Nick Buraglio
Energy Sciences Network; AS293
Lawrence Berkeley National Laboratory

+1 (510) 995-6068

On Thu, Apr 20, 2017 at 4:21 PM, David Farmer <> wrote:

For the ones I'm personally aware of, the issue is not provider related, they are using providers that will give them a full table if they asked for it.  In the cases I'm aware of it just the campus is using an high-end L2/3 switch or firewall capable of BGP routing, but not with a full global FIB or RIB for that matter, instead of a real router capable of holding 1M+ routes. However, I suppose that the lack of a real router could be compounded by provider issues too in some cases, but that hasn't been the case in the situations I'm aware of. 

On Tue, Apr 18, 2017 at 4:36 PM, Michael H Lambert <> wrote:

A question occurred to me just as we were hanging up on today's WG call.  It's directed at David, but others are more than welcome to chime in.  There are participants who have R&E tables from the regional and just point a default route at a transit provider.  Is it the case that some of these transit providers might be hard pressed, for various reasons, to send full tables to customers even if the customers could handle them?  I could very well see this happening with a mom and pop local cable company.

Michael

--

===============================================
David Farmer              
Networking & Telecommunication Services
Office of Information Technology
University of Minnesota  
2218 University Ave SE        Phone: 612-626-0815
Minneapolis, MN 55414-3029   Cell: 612-812-9952
===============================================