VT1000 and Comcast Speed Increase

Posts: 5 · New Forum Member Joined: Apr 18, 2005 Posts: 5

Just found out that Comcast is preparing to up our speeds again (Massachusetts)...who knows just WHEN that is going to be but still...

My current setup is: Cable Modem -> VT1000 -> Linksys WRT54G -> PC's

I have the VT1000 in front as for all my tweaking, I kept having issues with the QOS when it was behind the Linksys.

My question is will the VT1000 be able to handle the speed? We got a list of Cable Modems that won't handle in the increase in speed, and didn't know if this would also apply to the VT1000's. I did as Comcast, but they said to talk to Vonage, and after sitting on hold waiting for someone to understand what I meant, I've given up on that.

Is this going to be a factor, or should I be ok?

Posts: 646 · Posted: Wed Jul 13, 2005 2:48 pm Post subject:

You might have to experiment once the caps are raised. When I was on Purdue's campus network this last year, I noticed that people using routers usually couldn't achieve the full 100Mbit speeds. Some routers choked at around 5Mbit, others went up to 60Mbit or so. (I never kept track of which brands/models of routers they were)

So, a lot of routers can probably the handle the 8000/768 tier, but the best way to find out is to try it once it arrives.

Posts: 3198 · Posted: Wed Jul 13, 2005 4:40 pm Post subject:

and chances are the wrt will blow the vt1000 out of the water. of course that's true without speed upgrades, but if it ain't broke...

Posts: 646 · Posted: Wed Jul 13, 2005 4:49 pm Post subject:

I've always wondered whether the WRT54G could route at 100Mbit. Reebok, here's something you could try:

Connect one computer to a LAN port, another to the WAN port. Set up an FTP server or something, and get a direct file transfer going between the two. Measure the speed.

I think I'll try that with our RT31P2 tonight when everybody else is done with the Internet connection here. I've also got an old belkin router I could test...

Posts: 646 · Posted: Wed Jul 13, 2005 6:18 pm Post subject:

Ok, I've figured out a method for testing a router's throughput. I used the iperf program, available here:
http://dast.nlanr.net/Projects/Iperf/

Both of my test computers were Linux boxes, but it should work on Windows too.
(All subnet masks 255.255.255.0)

Router:
-------
LAN IP: 192.168.2.1
WAN IP: 192.168.3.1
WAN Gateway: 192.168.3.2
DMZ: 192.168.2.2

WAN PC:
-------
(connect with a crossover if the port isn't auto-sensing)
IP: 192.168.3.2
Gateway: 192.168.3.1
Command: "iperf -s"

LAN PC:
-------
IP: 192.168.2.2
Gateway: 192.168.2.1

Stats:
------
Belkin F5D5231-4:

Test upload, followed by download:
"iperf -c 192.168.3.2 -r"
upload: 8.05 Mbits/sec
download: 8.92 Mbits/sec

Test up & down simultaneously:
"iperf -c 192.168.3.2 -d"
upload: 4.28 Mbits/sec
download: 4.13 Mbits/sec

It looks like the crappy Belkin uses a 10Mbit half-duplex WAN port. I'll try the same test on my RT31P2 later tonight.

Posts: 646 · Posted: Wed Jul 13, 2005 10:05 pm Post subject:

As promised, here's the results for my RT31P2:

Sequential up/down:
upload: 26.3 Mbits/sec
download: 28.3 Mbits/sec

Simultaneous up/down:
upload: 16.2 Mbits/sec
download: 10.9 Mbits/sec

That's far from the capabilities of a 100Mbit ethernet port, but it's faster than any broadband available today, except for maybe high-end Fios service.

It'd be nice if people could rig up the same test with other routers, so we'd have a basis for comparison.

Posts: 646 · Posted: Fri Sep 02, 2005 2:07 am Post subject:

I just ran the same tests today through a WRT54G v2.2 running Sveasoft Alchemy 1.0, and the results were a bit underwhelming:

Sequential up/down:
upload: 24.6 Mbits/sec
download: 14.2 Mbits/sec

While the data was transferring, I had a telnet session open to the router, running "top". The CPU usage jumped up to 100% during the transfer, with the majority getting used up by "ksoftirqd_CPU0".

I was a bit surprised, but the WRT54G is actually slower than the RT31P2, in terms of LAN<->WAN bandwidth. I believe some of the later WRT54G versions have faster CPUs, so they might perform better than this one.

Maybe the RT31P2 just has less overhead because it's not running a full-blown Linux kernel?