[j-nsp] Question about Routing Engine Redundancy on MX

[Harry Reynolds]

My .02.

I too cannot find documentation that mandates the same RE hardware or Junos SW for all HA features. The 12.2 HA guide does mandate the same version for NSR, and my extension ISSU. (page 80):

http://www.juniper.net/techpubs/en_US/junos12.2/information-products/topic-collections/config-guide-high-availability/config-guide-high-availability.pdf

IMO, GRES/NSR/ISSU is complex enough. Generally speaking you are only adding potential for "less than desirable" outcomes if you attempt to deploy such features with mismatched RE hardware or Junos versions. It would be safe to say that testing HA features with all possible mismatches of RE HW and SW is not done routinely; it may work, or may not, and you may well be the first to know, being the first to ever try some specific test with some specific set of mismatched HW or SW.

As with all things in life, to some degree success or failure is a matter of how significant the mismatch is. Junos 7.0 on master and 12.2 on BU as opposed to 12.1R1 vs. 12.1R2; here the later will likely do OK... Same with RE hardware. If the master RE is significantly faster or has significantly more memory than the BU, you may hit timing issues with kernel or NSR synchronization, or, may find that post NSR things go poorly as the new master RE struggles to keep up now that it has to run the show. If you expect the BU RE to take over with no hit, it's quite reasonable to mandate that it at least have the same capabilities as the master.... 

In the end it's a complex interaction of how significant the mismatch is, whether it effects some key feature that is in use, and the degree to which the test box is scaled; in the end its intractable to try and predict all the possible permutations. 

As a final anecdote, I would never open a PR for a HA feature where the DUT had any such mismatch (unless it was a specific feature that mandated the disparity), as the first thing the engineer will ask is for me to try and repro with matched settings....

HTHs

Regards







-----Original Message-----
From: juniper-nsp-bounces at puck.nether.net [mailto:juniper-nsp-bounces at puck.nether.net] On Behalf Of Paul Stewart
Sent: Wednesday, January 09, 2013 10:52 AM
To: 'Jose Sanchez'
Cc: 'juniper-nsp'
Subject: Re: [j-nsp] Question about Routing Engine Redundancy on MX

Nothing that I have handy - sorry. we asked this question a few times to our Juniper SE and were told that (so presumed it to be factual)

 

J

 

From: Jose Sanchez [mailto:jasjuniper at gmail.com]
Sent: January-09-13 1:49 PM
To: Paul Stewart
Cc: juniper-nsp
Subject: Re: [j-nsp] Question about Routing Engine Redundancy on MX

 

Thank you,

 

Any link to the documentation that require this?

 

Thanks again

 

Jose

 

 

On Wed, Jan 9, 2013 at 12:43 PM, Paul Stewart <paul at paulstewart.org> wrote:

Both same hardware....


-----Original Message-----
From: juniper-nsp-bounces at puck.nether.net
[mailto:juniper-nsp-bounces at puck.nether.net] On Behalf Of Jose Sanchez
Sent: January-09-13 1:39 PM
To: juniper-nsp
Subject: [j-nsp] Question about Routing Engine Redundancy on MX

Hello,

Does anybody know if the RE Redundancy in Juniper MX Routers requires that both RE are the same hardware or it is enough that the REs has the same JUNOS Version?

Thanks

Jose

_______________________________________________
juniper-nsp mailing list juniper-nsp at puck.nether.net https://puck.nether.net/mailman/listinfo/juniper-nsp

_______________________________________________
juniper-nsp mailing list juniper-nsp at puck.nether.net https://puck.nether.net/mailman/listinfo/juniper-nsp

 

_______________________________________________
juniper-nsp mailing list juniper-nsp at puck.nether.net https://puck.nether.net/mailman/listinfo/juniper-nsp