weirdness with Extreme

Has anyone else seen this kind of thing before?

2 Extreme Black Diamond 6804s (ExtremeWare 7.2.0b37) at the core (active/standby) start to fight over which of them is master for a
particular VLAN. They go back and forth for maybe a minute, and for a brief period, the master switch seems to fight with itself over which will cede master status. The log messages indicate either high/low port count,
or high/low priority, for that particular vlan. ESRP then forces access switches hanging off the cores to flush their forwarding databases. All activity takes place within a span of just over a minute, then dies.

Extreme doesn't have anything to report yet. Just wondering if anyone has seen behavior like this in an Extreme environment.



We have a pair of cisco 7204VXR routers connecting to STFI receiving market data. At peak periods micro-bursts of unicast and multicast data overrun the Ethernet fifo buffer due to the 7200 being a cpu based router. A 7600 router would be a good replacement but it isn't cost effective. We need BGP, rip, pim multicast and netflow. Since the connections are all metro Ethernet, Cisco has suggested looking at the 3750 switch platform that does BGP since all of the packets are hardware switched, but it doesn't due L3 netflow. I've been doing cisco for too long and was wondering what the cost effective options are with other vendors or even other possible cisco solutions.

Hi Matthew,

The Juniper J6350 boxes are both cost effective and are claimed to do line 2Gbit/s of IMIX traffic I think.

We've several deployed between multitple DCs in Dublin and a load of J4350 at different layers. Stick 2GB of ram into each one and they'll go a long way.


Paul Kelly
Technical Director
Blacknight Internet Solutions ltd
Hosting, Colocation, Dedicated servers
IP Transit Services
Tel: +353 (0) 59 9183072
Lo-call: 1850 929 929
DDI: +353 (0) 59 9183091


Blacknight Internet Solutions Ltd,
Unit 12A,Barrowside Business Park,
Sleaty Road,

Company No.: 370845

We currently have 7204VXRs with the NGE-G1 controller. That's rated at 1Gbit/s and we aren't running that high yet. What we are running into is micro-bursts that overrun the ethernet's fifo buffer due to the cpu routing (even though it's CEF based and cisco has verified the configuration). We need something that is hardware switched such as the cisco 7600/6500 or the Foundry line as some have pointed out.

As far as our netflow requirements are, we need to be able to determine average line rates for different protocols/ and different multicast channels (not realtime). We use Fluke Network's NetFlow Tracker and it works with sFlow, IPFIX, Netflow and other netflow like protocols, so it looks like almost any of the hardware suggested will work.

What about a 6524 switch? Or a Juniper EX 4200?

I think a J series would be the way to go as well. Even the 4350 claims 1Gbps+ forwarding.

To give you an idea of cost, a J4350 will list about $5k J6350 will list about $10.5k. The 8-port GigE PIMs list at $1800 per, 16-port GigE PIM (dual height) at $3000. Of course, those are list prices...