[j-nsp] EX 4200 stability with BGP and OSPF redistribution ?

Dan Farrell danno at appliedi.net
Mon Jun 21 17:25:13 EDT 2010


We leverage the EX3200 and 4200's extensively in our network, for edge, core, and access.

As far as edge (ISP connectivity) we use EX3200's in pairs- each EX3200 has a separate peer session to each upstream provider, providing redundancy (high-availability) without merging the two units as one logical unit. This makes zero-downtime maintenance easier at your edge, as upgrading a stacked chassis involves rebooting all the devices at once. And they're cheaper than their 4200 counterparts.

I'm elated at the 4200's performance in our core- I think what may be of use to you is a comparison to equivalent Cisco gear- in this light we just replaced a two-chassis 3750G stack with a two-chassis EX4200 stack (we stack them to take advantage of port densities with staggered growth in the core), and we are glad we did so.

The EX series allows 1000 RVI's and 4k VLANS per virtual chassis- the Catalyst 3xxx series only actually supports 8 RVI's, and they don't publish this (you will find it when configuring the profile of the device). This created a problem with 10 OSPF interfaces (and 15 other non-OPSF interfaces) on the Cisco. Upon a link-state change on any of the Cisco's OSPF-configured interfaces, the CPU would crank up to 100% and the stacked device throughput was ground to a crawl (80%+ traffic loss). Changing the configuration in the OSPF subsection, elimination of the problem interface (flapping or not) from the configuration, or a complete reboot would solve the problem- none of which are attractive solutions to a problem we shouldn't have been having in the first place.

Compare this to a two-chassis EX4200-48T stack we have in another part of the network- 13 OSPF interfaces and ~845 other non-OSPF RVI's , and the stacked device hasn't given us any grief.  They cost us 1/3 less than the Cisco solution, and doubled the port density (the Ciscos had 24 and the Junipers we got have 48 ports).

There are platform limitations, like memory, which may cause you to be a little more exotic on BGP route selection, but the Catalyst 3750G's have even less memory as I recall. Overall they have been extremely good for our network, and have caused me to swear off Cisco completely.

Hope this provides some insight.

Dan

-----Original Message-----
From: juniper-nsp-bounces at puck.nether.net [mailto:juniper-nsp-bounces at puck.nether.net] On Behalf Of Laurent HENRY
Sent: Monday, June 21, 2010 6:29 AM
To: juniper-nsp at puck.nether.net
Subject: [j-nsp] EX 4200 stability with BGP and OSPF redistribution ?

Hi all,
        I am thinking about using two EX 4200 as redondant border routers of my main Internet link.

In this design, I would then need to use BGP with my ISP and OSPF for inside route redistribution.

Reading the archive, and on my own experience with the product too, i am looking for feedbacks about stability of this solution with EX.

In archives i understood there could have been some huge stability problems, am i right ?

Could things be different with 10.1 JunOS release ?

Does anyone actually use these features actively with this platform ?


Regards


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



More information about the juniper-nsp mailing list