[c-nsp] DMVPN scalability question on the 28XX ISR's

[Luan Nguyen]

I wouldn't say not recommended by Cisco though.  The DMVPN design guide is pretty old (2008) http://www.cisco.com/en/US/docs/solutions/Enterprise/WAN_and_MAN/DMVPN_3.html
I wish that Cisco would update that with ASR and ISR2 information and design guidance.  That's a very good document and the performance numbers are quite accurate.
When I first worked with DMVPN, most of the designs were dual hubs, dual cloud with EIGRP.  I was tempted with BGP as well, but mostly in a lab environment since operation folks didn't want to support it. 
Today, I believe the drive is toward single cloud, with tier layered...etc. 
I am using single cloud DMVPN design for a 3 hubs spoke-to-spoke TLS network with EIGRP and it has been working great.  Then again, the number of spokes is way < 2000.

-Luan


-----Original Message-----
From: Octavio Alvarez [mailto:alvarezp at alvarezp.ods.org] 
Sent: Wednesday, April 21, 2010 2:04 PM
To: Luan Nguyen; 'Engelhard'; rodunn at cisco.com; Erik Witkop
Cc: cisco-nsp at puck.nether.net
Subject: Re: [c-nsp] DMVPN scalability question on the 28XX ISR's

On Wed, 21 Apr 2010 06:35:37 -0700, Luan Nguyen <luan at netcraftsmen.net>  
wrote:

> In this case, a dual hub (loadshare/backup) for 1000+ spokes would be
> just fine.

Single-hub, dual-cloud scales and performs and converges better
than dual-hub, single-cloud and are not even recommended by Cisco.
Therefore, I would stick to the dynamic routing protocol approach.

-- 
Octavio.

__________ Information from ESET NOD32 Antivirus, version of virus signature database 5047 (20100421) __________

The message was checked by ESET NOD32 Antivirus.

http://www.eset.com


 

__________ Information from ESET NOD32 Antivirus, version of virus signature database 5047 (20100421) __________

The message was checked by ESET NOD32 Antivirus.

http://www.eset.com