Posted by: Anonymous Coward
on March 08, 2006 10:53 PM
Folks,
Hello folks,
I am a professional network and systems engineer who works with network infrastructures every day, mostly Cisco and (a little) Juniper. When I first learned about Vyatta, I was very excited, as I hoped to see an actual router product, using entirely FOSS, that can compete with Cisco's offerings.
Upon looking at this more, though, I don't have a lot of faith that, by itself, it'll have any effect. It's just another CD version of Freesco with Asterix added, something that "out of the box" thinkers have been doing for years with GNU/Linux and, for that matter, FreeBSD. It's just another Debian distro in a sea of distros.
What they're going to need is to be able to do like Damn Small Linux does and run from some kind of flash memory instead of a CD. Something not identical to, but along the lines of, this:
Add the optional 512MB IDE Flash Card, and you have a $400 box that 1.) doesn't use any fans, 2.) doesn't have any moving parts, and 3.) has the processing and DRAM oomph to do what a router needs to do (and then some!). But you'd certainly need a second network interface. The notion of depending on a hard disk or CD-ROM drive in a router is one that we network folks generally laugh at; they're much more prone to fail than flash memory is.
To compete with Cisco's routers (e. g. the 2800 and 3700/3800 series), and definitely any of Juniper's (I'm thinking the M7i, their smallest box, here), you'd need to have--at a minimum--the capability for four interfaces on it, including the option for, at a minimum, the following types of interfaces today:
1.) Frame Relay, up to and including T-3 speeds; 2.) ATM, up to and including OC-48 speeds; 3.) SONET, up to and including OC-48 speeds; 4.) 10/100/1000Mbps copper-based Ethernet; and 5.) 1000Mbps fiber-based Ethernet.
These interfaces would need to be hot-swappable, at least with the same type (Frame Relay T1 for Frame Relay T1, for example), in case an interface goes bad. Also, if you're going to go up against Juniper, the box would need to be able to support the max # of interfaces--the fastest interfaces, which is currently Gig-E--at full interface speed. Juniper's M7i can do this.
You'll also need something like Cisco's policy-based routing, which is how we send traffic to our Web content filter servers; Juniper has something similar. Add to that CoS and QoS, which you have to have for not just IP telephony, but quite a few other situations. Finally--and anyone who's ever used Symantec Ghost on a routed network can appreciate the need for this--you'd need multicast routing support.
In short, Vyatta would need to market and support an actual turnkey box that has feature parity with what I do with a router now. In its current form, all but a very few enterprise network managers will, sadly, tell Vyatta to go jump in the lake. Sad, but true.
Learn about Linux
Download Linux
Get Linux help
I don't see this going far, unfortunately
Posted by: Anonymous Coward on March 08, 2006 10:53 PMHello folks,
I am a professional network and systems engineer who works with network infrastructures every day, mostly Cisco and (a little) Juniper. When I first learned about Vyatta, I was very excited, as I hoped to see an actual router product, using entirely FOSS, that can compete with Cisco's offerings.
Upon looking at this more, though, I don't have a lot of faith that, by itself, it'll have any effect. It's just another CD version of Freesco with Asterix added, something that "out of the box" thinkers have been doing for years with GNU/Linux and, for that matter, FreeBSD. It's just another Debian distro in a sea of distros.
What they're going to need is to be able to do like Damn Small Linux does and run from some kind of flash memory instead of a CD. Something not identical to, but along the lines of, this:
<a href="http://www.damnsmalllinux.org/store/Mini_ITX_Systems/Damn_Small_Machine" title="damnsmalllinux.org">http://www.damnsmalllinux.org/store/Mini_ITX_Syst<nobr>e<wbr></nobr> ms/Damn_Small_Machine</a damnsmalllinux.org>
Add the optional 512MB IDE Flash Card, and you have a $400 box that 1.) doesn't use any fans, 2.) doesn't have any moving parts, and 3.) has the processing and DRAM oomph to do what a router needs to do (and then some!). But you'd certainly need a second network interface. The notion of depending on a hard disk or CD-ROM drive in a router is one that we network folks generally laugh at; they're much more prone to fail than flash memory is.
To compete with Cisco's routers (e. g. the 2800 and 3700/3800 series), and definitely any of Juniper's (I'm thinking the M7i, their smallest box, here), you'd need to have--at a minimum--the capability for four interfaces on it, including the option for, at a minimum, the following types of interfaces today:
1.) Frame Relay, up to and including T-3 speeds;
2.) ATM, up to and including OC-48 speeds;
3.) SONET, up to and including OC-48 speeds;
4.) 10/100/1000Mbps copper-based Ethernet; and
5.) 1000Mbps fiber-based Ethernet.
These interfaces would need to be hot-swappable, at least with the same type (Frame Relay T1 for Frame Relay T1, for example), in case an interface goes bad. Also, if you're going to go up against Juniper, the box would need to be able to support the max # of interfaces--the fastest interfaces, which is currently Gig-E--at full interface speed. Juniper's M7i can do this.
You'll also need something like Cisco's policy-based routing, which is how we send traffic to our Web content filter servers; Juniper has something similar. Add to that CoS and QoS, which you have to have for not just IP telephony, but quite a few other situations. Finally--and anyone who's ever used Symantec Ghost on a routed network can appreciate the need for this--you'd need multicast routing support.
In short, Vyatta would need to market and support an actual turnkey box that has feature parity with what I do with a router now. In its current form, all but a very few enterprise network managers will, sadly, tell Vyatta to go jump in the lake. Sad, but true.
#