<div dir="ltr">Hi All,<div><br></div><div>Like Basile we use a number of MLX-16 and MLX-8's in our core SP network. We are using a the router as a MESR with full IPv4 and IPv6 feeds along with Multicast routing (PIM-SM in a VRF for IPTV) with MPLS (primarily VLL and VPLS, no IPVPN yet) for metro Ethernet services and DSL backhaul.</div>
<div><br></div><div>Depending on your scaling requirements such as:</div><div>- number of BGP peers</div><div>- IPv4, IPv6, IPVPN routes</div><div>you may need to look at the XMR. Currently we have 15 full IPv4 peers and 1 full IPv6 feed on one of our route-reflectors and it's operating at 65% memory. Your milage may vary.</div>
<div><br></div><div>I would recommend the MLX-8's and MLX-16's as they use common parts (fabric and power, except fans) whereas the MLX-4 uses different fabric and power to the larger chassis. This makes it easier for sparing, and the difference in cost of the chassis is not significant....</div>
<div><br></div><div>In response to Basile, there is a new multi-service CAM (aptly named multi-service-2) profile on 3.8 which does allow full IPv4 table (up to 390k), full IPv6 table along with a reasonable allocation of VPLS, VLL and IPVPN instances.</div>
<div><br></div><div>Like Basile we are using both 1G and 10G interfaces and also receive great support directly from Foundry. We have found the stability of MLX platform to be good.</div><div><br></div><div>If you have any specific requirements or scaling questions I would highly recommend talking to an SE.</div>
<div><br></div><div>If you really want (arguably) less value for money, you could also look at:</div><div>- Cisco 76xx</div><div>- Alcatel 7450/7750</div><div>The Juniper MX and Foundry MLX/XMR use different architectures, understanding these differences helps with the determining the pros and cons, but they currently have similar scaling numbers (XMR) in routes, the MX might have an edge on number of peers. </div>
<div><br></div><div>Kind Regards,</div><div>C.</div><div><div><br><div class="gmail_quote">On Tue, Sep 2, 2008 at 5:25 PM, Basile Bluntschli <span dir="ltr"><<a href="mailto:basile.bluntschli@gmail.com">basile.bluntschli@gmail.com</a>></span> wrote:<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex;"><div dir="ltr">Hi Samit<div class="Ih2E3d"><br><div class="gmail_quote"><blockquote class="gmail_quote" style="border-left:1px solid rgb(204, 204, 204);margin:0pt 0pt 0pt 0.8ex;padding-left:1ex">
<br>
I am looking for a L2/L3 capable ethernet router with 20Gig ports in my<br>
core, that should do full bgp feeds from multiple upstreams w/ , IPv6<br>
and multicast routing, MPLS and 4byte Asn in future. I am currently<br>
looking into 4 products from 4 vendor.<br>
<br>
1. Extreme BD 1280xR<br>
2. Force10 E300 (don't have MPLS but might do in future)<br>
3. Foundry NetIron MLX-4<br>
4. Juniper Mx240.<br>
<br>
Suggestion:<br>
<br>
1. Price the killer<br>
2. Stability and reliability<br>
4. Performance<br>
3. Support<br>
</blockquote></div></div>We use all Foundry now in our core and they are really reliable L2/L3 device. As i don't know the other products i can tell you just my opinion about the Foundry MLX-4. <br><br>2. We have some MLX-4 but we are not that happy with those. They, for the first view, offer almost the same specifications as the XMR product range does. But in fact there are quit some difference. <br>
<br>For example if you have a look at the CAM Partitioning Profiles you see that the XMR offers you more services at the same time. What i read out of your post the only usable profile on a MLX would be the ipv4-ipv6-2 which allows you to use the BGP fullfeed IPv4 and the BGP fullfeed IPv6. This profile does not support L3VPN, but you can do VLL and VPLS.<br>
<br>Then we suffered some problems on the MLX with the memory usage when having a lot of BGP sessions configured. We used the MLX as a route-reflector in our core for our edge devices. This used all of the available memory (1Gig) on the MLX, which lead to a device who was not configurable anymore (conf t was rejected because of insufficient memory). We then had to change the MLX with a XMR (which has 2Gig RAM) as upgrades of the Ironware (IOS) did not help. In future release there should be a fix for this, foundry told us. If you have a look at the new release notes for the 3.9 Ironware there seems to be improvement for this issue, so maybe in future the MLX will also be capable of this...<br>
<br>4. But in fact also during this memory problem the device was stable and performance was never an issue. We use the MLX and XMR with 10Gig Interfaces and they provide line rate routing.<br><br>3. We have a direct contact at foundry where we get great support. <br>
<br>To make the story short my opinion is to use an XMR if you really want to have it as a core device. MLX is fine for edge purposes.<br><br>Kind regards<br>Basile<br><br></div>
<br>_______________________________________________<br>
foundry-nsp mailing list<br>
<a href="mailto:foundry-nsp@puck.nether.net">foundry-nsp@puck.nether.net</a><br>
<a href="http://puck.nether.net/mailman/listinfo/foundry-nsp" target="_blank">http://puck.nether.net/mailman/listinfo/foundry-nsp</a><br></blockquote></div><br></div></div></div>