[j-nsp] Seeking Input on Rough Scaling for T and M series MPLS PE Terminations

[truman at suspicious.org]

Hi Mark,

Exactly; scaling is always a difficult topic to clearly articulate. Sometimes 2+2=5.

Cheers,
Truman


Sent via BlackBerry® from Telstra

-----Original Message-----
From: Mark Tinka <mtinka at globaltransit.net>

Date: Tue, 21 Apr 2009 13:50:18 
To: <juniper-nsp at puck.nether.net>
Cc: Truman Boyes<truman at suspicious.org>; Loopback EZ<loopback at ezxyz.com>
Subject: Re: [j-nsp] Seeking Input on Rough Scaling for T and M series MPLS PE Terminations


On Tuesday 21 April 2009 01:38:01 pm Truman Boyes wrote:

> Depending upon your configurations, the actual scaling
> figures may vary. Here are some figures that should help
> in rudimentary sizing: you can setup 2000 VPLS instances
> w/ 100k MAC addresses;  you can setup 32k CCC
> connections, and somewhere between 2000 -  6000 routing
> (VRF / virtual routers) instances.
>
> BGP peers, you should be able to get more than 2000+ ...
> and possibly up to 4k depending on how many routes you
> are exchanging with your peers.

I imagine these figures depend on whether the router is 
performing just one of the above functions you mention, or 
more/all of them, i.e., the scalability numbers of each 
service would vary depending on what other competing 
features/services are being enabled at scale as well, no?

Cheers,

Mark.