[f-nsp] NetIron MLX-4 vs Juniper MX240

[Brad Fleming]

> On the MLX side of things.  With a rather large Foundry switching  
> environment, I and my team are very comfortable on that platform.   
> The switches just work, and the strangest problem I have seen  
> interop problems with a Cisco -- and I blame Cisco for that.  We  
> have had some struggles using the Foundry ServerIron due to a few  
> bugs here and there.  I do, however, expect that a full layer3 stack  
> is significantly less complicated of code than what the ServerIron  
> is able to do, so should have less bugs.

This is my experience as well. Other than a couple buys involving (1)  
QoS + MPLS, (2) adding/removing VPLS's too quickly, and (3) the BGP AS  
path length problem we've not had any issues. Both of those bugs were  
already identified and resolved.. we just had to do a quick software  
upgrade. Since we're a research network, we use quite a few of the  
features.. and most of the newer features.
>
> There is a cost difference, but it's not even 25% for our config.

Really? I've not priced an MX device recently but to get a box with *- 
R cards used to be stupidly expensive. Then if you wanted to do  
advanced QoS it got completely insane.
>
> Last but not least, support.  Both vendors have terrible technical  
> support when you have a bug, in my experience.  I loathe having to  
> open a case.  Is it different for support on the router series?

We were early SRX adopters as well. Painful indeed. Hopefully you are  
not trying to manage the devices via NSM. :(

I've experienced pretty solid support from Brocade since the merger/ 
buyout/whatever. Juniper's support at the SRX and J-Series level is  
pretty weak IMHO. Things like "we don't have IPv6 in the lab so can  
you mock this up for us?" is not what I want to hear from a router  
manufacturer. I'm sure the MX support engineers are better though.



I like that the MLX has a high density 10G card available (hoping the  
XMR gets one soon). I don't know if total throughput is a major  
concern for you, but an MLX 4000 with four 8x10G cards represents a  
significant amount of throughput in a very small form factor. And I  
**believe** the MLX has a roadmap to 100G, but be sure to check with  
your SE to be sure.

Next to last note: I've not seen specs on the MX recently but the MLX/ 
XMR are very efficient with space and power. We were surprised how  
little the XMR with a couple 4x10G cards consumed. Certainly makes  
leasing space and power in someone else's facility easier!

A final note: We perform software upgrades once a year (during the  
summer) unless there's an emergency. Having a core router that can  
reload, take multiple full table BGP feeds, converge everything and be  
routing traffic 3 minutes after hitting enter at the "reload" command  
is really nice. We have so much confidence in the devices, we reload  
all of them at the exact same time.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://puck.nether.net/pipermail/foundry-nsp/attachments/20100507/eb12a5aa/attachment.html>