January 2004 - Test - lists.nanog.org

Re: PC Routers (was Re: /24s run amuck)
by james 15 Jan '04

15 Jan '04

: It seemed that zebra was not following the RFC for OSPF. This would be one advantage to Quagga over Zebra. It is my understanding there have been many changes in Quagga to OSPF to make it standards compliant. James Edwards Routing and Security jamesh(a)cybermesa.com At the Santa Fe Office: Internet at Cyber Mesa Store hours: 9-6 Monday through Friday 505-988-9200 SIP:1(747)669-1965

1 0

Re: PC Routers (was Re: /24s run amuck)
by Randy Bush 15 Jan '04

15 Jan '04

> traffic doubled and tripled in a year, it didn't go 10x. actually, at the time, mo said doubled every nine months. and it did. randy

1 0

RE: PC Routers (was Re: /24s run amuck)
by Michel Py 15 Jan '04

15 Jan '04

>> almost all times I hear people saying there is problem >> with Zebra or Quagga, more than half of all times it >> is problem with their OS, not the daemon. > and we care because? the router is a black box. if the > output is not what is expected, it matters not why. > though understandable, it is still not acceptable. The main issues I have with zebra are: 1. The need to install an OS on the host. 2. The need to harden it. 3. The possible hard disk failure (having *nix on ATA flash is no better given the actual limits in the number of times one can write to flash). There are things that I don't like with Cisco, but one thing I do like is that it boots from flash and it takes no time to install an image, remove the pcmcia card from the router, and boot different images from the flash with the flip of a config command. The concept of appliance (vs. computer) comes to mind. That being said, How does zebra deal with QOS/priority/custom/queuing/LLQ? With CAR? With IDS? With route redistribution to/from OSPF or ISIS? With multichassis multilink PPP? With spanning tree on multiple VLANs? With peer groups? With SNMP? How does the host deal with 802.1q trunks? With Channel interfaces? With hot-swapping a line card? With TCP MD5? These are the questions I ask myself when I pick a routing platform. Cheap is of no use to me if it does not do what I need. Michel.

5 8

Re: Looking for power metering equipment...
by Mark E. Mallett 15 Jan '04

15 Jan '04

On Thu, Jan 15, 2004 at 11:40:54AM -0500, doug(a)nanog.con.com wrote: > Do you know a model number? I can't seem to find anything like this on > radioshack.com. (cc'd to nanog ..) Shoot, I should have looked first. I can't find it either. I found the note from January 2003 where I heard about it, and it said: http://www.radioshack.com/product.asp?catalog%5Fname=CTLG&category%5Fname=C… or just go to radioshack.com and search for watt meter (two words) under test equipment orwhatever.. it says they're sold out online, so I don't know if they discontinued it after not getting a lot of sales. The last sentence is foreboding. Sorry about that. mm

3 2

Re: PC Routers (was Re: /24s run amuck)
by james 15 Jan '04

15 Jan '04

: Be real carfull with Zebra, I got stung big time !!! What I run is actually Quagga, despite saying Zebra. Would you share your experiences ? James Edwards Routing and Security jamesh(a)cybermesa.com At the Santa Fe Office: Internet at Cyber Mesa Store hours: 9-6 Monday through Friday 505-988-9200 SIP:1(747)669-1965

8 11

Looking for power metering equipment...
by Alex Rubenstein 15 Jan '04

15 Jan '04

Preamble: We run a colocation center. We sell power to customers. Question: We are looking for something that sits in the PDUs or branch circuit-breaker distribution load centers, that, on a branch-circuit by branch-circuit basis, can monitor amperage, and be queried by SNMP. Considering there are several hundreds of circuits to be monitored, cheap and featureless (all we need is amperage via SNMP) is fine. Looked at things like Square-D PowerLogin stuff, but thats very pricey, and does about 30x what we need. Pointers? URLs? Experiences? Thanks.

6 6

RE: PC Routers (was Re: /24s run amuck)
by Michel Py 15 Jan '04

15 Jan '04

> Deepak Jain wrote: > With a network boot OS for each POP, you can do > version control much much more easily. This is seriously flawed, IMHO. I'd encourage my competitors to do it: after the master image gets corrupted all it takes is a bozo tripping the right circuit breaker and the entire POP is kaput. > QOS, priority/custom queueing are all KERNEL/underlying > OS functions. This also is flawed, IMHO. What if you want to do queing or QOS based on BGP? Michel.

2 1

RE: PC Routers (was Re: /24s run amuck)
by Michel Py 15 Jan '04

15 Jan '04

> Alexei Roudnev wrote: > Purchase SuperMicro U1 server, with 2 9 Gb SCSI > disks (hot swappable). Suddenly that cheap router ain't cheap anymore. > Now, say, announce A crash Cisco IOS. 99.9% Internet backbones > are Ciscos, so this announce breaks few Ciscos around and die > - so you never know about it (and will not have a chance to be > happy that _this announce crash Ciscos but do not crash ZEBRA). > Not bad, of course - you are alive, all Internet is alive. > Now, say, announce B crash ZEBRA (and do not crash Cisco). It > will spread until it reach first ZEBRA on it;'s road - _your_ > ZEBRA. So all Zerbras in Internet crash at once (and you are > unhappy). Another variant: announcement C crashes vendor "A", but not vendor "B" and not Zebras either (put whoever you want for "A" and "B" but there's only two of them on the backbone, mostly). Unfortunately, it takes a few minutes to crash, so it has enough time to propagate all over the Internet before the first "A"s begin to crash. As more "A"s crash "B"s will quickly be overwhelmed and the entire internet soon is down because no matter who coded it, when half of the backbones takes a hike the other half follows. Your Zebra is still up, but it does not do you any good because the entire Internet is down and you're only a small leaf on its edge, so nobody knows that you are still up, and they don't care anyway because no matter what they can't get anywhere. And yes this happens; not to beat or pick on any of the parties, I remember the entire AT&T frame relay network being down nationwide for more than 24 hours (for parts) because (as it is rumored) someone pushed a bad piece of software on a Stratacom switch. [me puts the asbestos suit on] <flame_bait> <FUD> If you have vendor C or vendor J, and all vendor C or J routers crap out at the same time, you're safe. Yes, you were down but so was half of the rest of the world, so it's obviously not your fault but vendor C or J's fault. If you have a zebra one a homebrew PC and all Zebras crap out: 1. You can't blame C or J. Worse, you can't blame anybody else. 2. As a coincidence, a sales droid from C or J will see your boss in the following days, take him/her to a nice restaurant, and will inevitably say "this would not have happened if you had a real router instead of this garage-built crud". 3. The sales droid is full of it, but _you_ are deep into it. </FUD> </flame_bait> Life is not fair, is it? Michel.

2 1

Re: PC Routers (was Re: /24s run amuck)
by Michael.Dillon＠radianz.com 15 Jan '04

15 Jan '04

>If someone were to take *half* the software innovations which have been >made over the past 15 years (a decent fib, interrupt coalescing, compiled >packet matching rulesets, etc) and applied them as if they knew something >about networking and coding, they could very easily produce a box using >off the shelf PC hardware which woops up on a 7206vxr for somewhere less >than $2000. Do you have any evidence that these improvements are not being done? The people building supercomputer arrays using Linux have a need for consistently high pps and bps that is greater than anything we see today on the Internet. They've been working on these types of improvements in device drivers and the OS (Linux, *BSD) for years now. You might not find this stuff in a standard enterprise distro like RedHat or SUSE but it is trvial to source this stuff and integrate it into your own build of the OS. A lot of this discussion has been people guessing about performance issues but few people have taken the time to put together a few boxes with Linux or *BSD and either Zebra or Quagga to trial them. We all go through detailed evaluations when buying C or J boxes, so it's not a waste of time to trial some Z or Q boxes as well to see what they can do. In the end, the resulting performance is affected by so many factors that it can't be predicted without testing. For instance, any weaknesses in the software might be completely nullified by the greater CPU power of a PC platform. And lets not forget that there are other platforms like ARM and PPC. Here's an ARM development system with PCI http://www.simtec.co.uk/products/EB110ATX/intro.html and here's a PPC one http://www.artesyncp.com/products/PM-PPC-440.html And if anyone thinks that ASICs give C and J an speed advantage that others can't touch, then guess again. Nowadays those ASICs are probably programmable ASICs which is a fancy way of saying that they are mostly made up of FPGA cells. It is not that difficult or expensive for people to design and build their own ASIC using cheap FPGA technology from companies like Xilinx. This is basic sophomore level electronics and is simple and cheap enough that people even hack their own MP3 players using FPGAs http://www.pjrc.com/tech/mp3/fpga/ PC-based routers may not be magic bullets but I think we should take them a lot more seriously especially if you want to innovate and offer something that differentiates you from other network operators. In a world where everybody runs C and J networks, there is only one flavor available, vanilla. --Michael Dillon

1 0

Re: PC Routers (was Re: /24s run amuck)
by neil＠DOMINO.ORG 15 Jan '04

15 Jan '04

> This also is flawed, IMHO. What if you want to do queing or QOS based on > BGP? That doesn't make any sense. You could only do the signalling for such a requirement in BGP and that isn't too hard to implement but the actual work to do QoS/queuing are in the kernel/OS/architecture irrespective of vendor or platform. Regards, Neil.

1 0