[j-nsp] anyone running VC with 2 * EX4500?

Ben Dale bdale at comlinx.com.au
Tue Feb 21 17:07:51 EST 2012


Hi Alex,

On 21/02/2012, at 8:38 PM, Alexander Bochmann wrote:

> we've been putting off converting our EX4500s to a virtual 
> chassis for quite some time now. I've seen a few posts about 
> mixed EX4500/4200 setups, but none with several EX4500s. 
> 
> Does anyone run something like that? Any special caveats, 
> and should we wait some more before trying it? 

I've just finishing a deployment with a core of 4x EX4500s in a single VC (spread across two locations 27km apart), with an "access" tier consisting of several VCs, each containing mixed EX4500/EX4200 in various combinations (4x 4500 and 2x 4200 mostly).

I'll echo Phil's statements about needing a few extra reboots when bringing it up for the first time, but this is mostly due to the requirement for mixed-mode (which requires a reboot).  Since then it has been fairly solid.  

I'm running 11.4R1 across the board, but looking at the release notes for 12.1, there seems to be a lot more features added to smooth the software upgrade process (selective rebooting of nodes, quasi ISSU etc).  Just not enough for me to delay this deployment any further.

I've been testing Non-Stop Bridging pretty extensively, and it works as advertised - virtually hitless L2 fail-over when you kill the power to the master RE, and no xSTP re-convergence.

On the other hand Non-Stop Routing is currently exhibiting a weird issue whereby failing from one RE to another is fast (sub-second), but failing back the other way takes about 5 seconds.  There doesn't seem to be any reason for this happening and it's always slower in the same direction even after a reboot.  I can live with this for the moment, but it's a bit sad that fail-over is now slower than HS/VRRP which this is replacing.  Guess I've got a date with JTAC in the near future.

Cheers,

Ben


More information about the juniper-nsp mailing list