[c-nsp] Router/switch recommendations for colocation

[Michael K. Smith - Adhost]

Personally?  I would look at your favorite *nix and Bird, but that's just me.  J  The 2900 also looks to fit well in this niche.  I'm also a big fan of separating my edge routing from my core, so I would look at a "router" like the 2900 for the edge and then lighter L3 intelligence in the core.

Mike

--
Michael K. Smith - CISSP, GSEC, GISP
Chief Technical Officer - Adhost Internet LLC mksmith at adhost.com<mailto:mksmith at adhost.com>
w: +1 (206) 404-9500 f: +1 (206) 404-9050
PGP: B49A DDF5 8611 27F3  08B9 84BB E61E 38C0 (Key ID: 0x9A96777D)

From: David [mailto:webnetwiz at gmail.com]
Sent: Monday, January 31, 2011 11:02 AM
To: Michael K. Smith - Adhost
Cc: cisco-nsp at puck.nether.net
Subject: Re: [c-nsp] Router/switch recommendations for colocation

I'll agree that folks certainly don't follow the BGP-advertisement best-practices, but some of the stuff you see in IPv6 advertisements today are still folks experimenting with the protocol. ISPs themselves need to police this in regards to how they allow their customers to peer with them, and have stringent requirements for the advertisements they receive from the customers. Most Tier1 SPs in the US do that already, and the smaller guys need to follow suit. Majority of people polluting the table are operating networks outside of the US. And such productive discussions about the state of the global table, and what needs to be done, etc. are good to have at places like NANOG, but customers have needs today and budgets are still tight, and that's one of the reasons enterprises hold off on deploying or thinking about IPv6 (a bad practice for sure). But again I come back to the original goal of the post, two IPv4 sessions, no IPv6 (at least yet). The poster isn't interested in the state of what's going on with the Internet at large, rather what piece of equipment to use to solve this particular technical requirement. I am interested in what you think is the proper piece of equipment that best fits here.

David.
On Mon, Jan 31, 2011 at 10:28 AM, Michael K. Smith - Adhost <mksmith at adhost.com<mailto:mksmith at adhost.com>> wrote:
Not peer via IPv6?  Really?  And, given what we have experienced with the v4 routing table bloat, I can't imagine that moving to v6 will change operators' tendencies to deaggregate.  It just means we have more space to do it.    If you check the v6 table today, you will already see lots of more-specific announcements, /64's (depending on your provider), etc. that shouldn't be there.  I haven't heard anyone say that v6 was going to help the DFZ keep from expanding.  Quite the contrary.

Mike

From: David [mailto:webnetwiz at gmail.com<mailto:webnetwiz at gmail.com>]
Sent: Monday, January 31, 2011 10:22 AM
To: Michael K. Smith - Adhost
Cc: cisco-nsp at puck.nether.net<mailto:cisco-nsp at puck.nether.net>

Subject: Re: [c-nsp] Router/switch recommendations for colocation

Why would there be a need to forklift? If the box will not do peering via IPv6, then modify the CAM profile to allocate more memory for IPv4, and that takes care of 500+K IPv4 routes. The key thing here is how many routes are held in the FIB, which holds the best routes after BGP machine runs through its calculations, and as long as the box can hold all of the peer's routes in the RIB-in, then you're ok. Hopefully when we allocate all /8 this year, folks will start moving over to IPv6. With IPv6's aggregation capabilities, you won't need such a large table, so you're ok there. The OP is only doing two BGP sessions, and it seems requires a multitude of GigE ports, so an ASR here gets expensive fast for this type of deployment. I am curious though Mike... what box would you suggest here?

David.
On Mon, Jan 31, 2011 at 9:24 AM, Michael K. Smith - Adhost <mksmith at adhost.com<mailto:mksmith at adhost.com>> wrote:

> -----Original Message-----
> From: cisco-nsp-bounces at puck.nether.net<mailto:cisco-nsp-bounces at puck.nether.net> [mailto:cisco-nsp-<mailto:cisco-nsp->
> bounces at puck.nether.net<mailto:bounces at puck.nether.net>] On Behalf Of David Kotlerewsky
> Sent: Monday, January 31, 2011 9:08 AM
> To: cisco-nsp at puck.nether.net<mailto:cisco-nsp at puck.nether.net>
> Subject: Re: [c-nsp] Router/switch recommendations for colocation
>
> Sooo, does it have to be Cisco???
>
> I am currently helping a customer with a very similar situation, and I think
> based on what you're trying to do here, you should look outside of what
> Cisco can offer you. In my mind this is a perfect fit for a Brocade NetIron
> CER 2000 switch/router. It can take in multiple BGP feeds, it has MPLS
> features if needed, oh, and it's only a 1U box with RPSUs. Also, depending
> in the actual model you choose to go with, I think you should be able to get
> it cheaper than an ASR, and support costs are lower. Cisco 37XX switches are
> really not data center grade for lack of memory/buffer space and MPLS/VPLS
> features. I have customers running NetIrons in their cores as well as PoP PE
> routers, and everything works as its supposed to.
>
> Just my $0.02
>
> David.
>From the literature:

The NetIron CER 2000 can store up to
512,000 IPv4 or 128,000 IPv6 unicast
routes, enough to accommodate the full
IPv4 Internet routing table today and
provide a smooth migration path to IPv6.

That is not going to hold you for very long, IMO.  If you buy this box for BGP you are going to be doing a forklift upgrade in less than two years.

Mike