Cable internet, and bike headlight
Nov. 28th, 2005 07:08 am![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png){kind=small}
johnridleyOur cable connection continues to be a little unreliable, but it turns out to be equipment at our end.
We've been having to reset our equipment for quite a long time, but moreso in the last several months. I eventually figured out that the chance of loss of connection is proportional to the number of connections we have going, and almost all of those are from P2P apps on my machine. I've been using them more heavily in the last 6 months or so
It made sense to me that the router would be the problem. I've been playing with different firmwares for our Linksys WRT54G. Most of the time when the connection went down, I couldn't get to its web interface. A lot of the other times when the connection WAS running but it was heavily labored, pulling up the web interface was very slow, like 45 seconds to get the "enter password" dialog. Well, it is pretty old, a version 1.1 (current is 5.0) and has a pretty slow CPU.
So about mid-November I went down and picked up a Netgear wireless-G router. It's working a treat. I can always get to it and it seems fast.
However, the connection problems continue. But now rather than resetting BOTH boxes, I only need to reset the modem. When the problems are happening, I can get to the router's web interface, but not the modem's at 192.168.100.1 Now, I don't really understand the guts of what the modem is doing; I don't know if it needs to do extra work per connection, like a router, or whether it's just passing bytes, like a modem. I know purists itch when you call it a modem, so maybe it's doing more than passing bytes.
Well, anyway, we paid $50 for the modem, and have used it for 3 years, so it's more than offset the $5/month that Charter was charging us for renting a modem. I started shopping around, and I found user comments on EVERY current modem brand and model saying that it couldn't handle large numbers of P2P connections. Finally I gave up and bought a Motorola SB5120, to replace the Motorola SB4200 we're running now. It should be here later in the week. I'll report back. All the modems cost $65 to $80, so you might as well get one that works well. The Motorola is DOCSYS 2.0 compliant and has a fast CPU, so it should be good for a while.
In other news, the HID bicycle headlight was repaired by the manufacturer, I modified a switching voltage regulator I build years ago to work with it, set it at 12.8 volts, and it's working fine now. Hopefully it'll keep burning for a few years at least.
We've been having to reset our equipment for quite a long time, but moreso in the last several months. I eventually figured out that the chance of loss of connection is proportional to the number of connections we have going, and almost all of those are from P2P apps on my machine. I've been using them more heavily in the last 6 months or so
It made sense to me that the router would be the problem. I've been playing with different firmwares for our Linksys WRT54G. Most of the time when the connection went down, I couldn't get to its web interface. A lot of the other times when the connection WAS running but it was heavily labored, pulling up the web interface was very slow, like 45 seconds to get the "enter password" dialog. Well, it is pretty old, a version 1.1 (current is 5.0) and has a pretty slow CPU.
So about mid-November I went down and picked up a Netgear wireless-G router. It's working a treat. I can always get to it and it seems fast.
However, the connection problems continue. But now rather than resetting BOTH boxes, I only need to reset the modem. When the problems are happening, I can get to the router's web interface, but not the modem's at 192.168.100.1 Now, I don't really understand the guts of what the modem is doing; I don't know if it needs to do extra work per connection, like a router, or whether it's just passing bytes, like a modem. I know purists itch when you call it a modem, so maybe it's doing more than passing bytes.
Well, anyway, we paid $50 for the modem, and have used it for 3 years, so it's more than offset the $5/month that Charter was charging us for renting a modem. I started shopping around, and I found user comments on EVERY current modem brand and model saying that it couldn't handle large numbers of P2P connections. Finally I gave up and bought a Motorola SB5120, to replace the Motorola SB4200 we're running now. It should be here later in the week. I'll report back. All the modems cost $65 to $80, so you might as well get one that works well. The Motorola is DOCSYS 2.0 compliant and has a fast CPU, so it should be good for a while.
In other news, the HID bicycle headlight was repaired by the manufacturer, I modified a switching voltage regulator I build years ago to work with it, set it at 12.8 volts, and it's working fine now. Hopefully it'll keep burning for a few years at least.
![[identity profile]](https://www.dreamwidth.org/img/silk/identity/openid.png){kind=small}
no subject
Date: 2005-11-28 04:34 pm (UTC)Yeah, it's definitely doing more.
At this point, I'm fairly certain that modems sold by the cableco and telcos are doing various forms of bandwidth shaping. It wouldn't surprise me if what you're running into is put in by design.
I remember reading something very recently dealing with an ISPs BW shaping models causing random weird observed side effects similar to yours. Of course, I can't remember where I saw this. If I do, I'll forward it.
no subject
Date: 2005-11-28 04:54 pm (UTC)What I'm seeing is actually a sudden, complete and persistent (lasts until power cycle) death of all traffic in both directions when the modem decides to go titsup. I can't even get to the modem's built-in web server when it's in this state. I hope that isn't by design.
no subject
Date: 2005-11-29 05:12 pm (UTC)Are you sure you mean ICMP here? It'd be weird to use that for anything related to protocol rate limiting, unless someone is insane enough to use source quench.
If you've got a SW router in front of your connection, you might be better off setting up some in-house QoS so that you don't end up in this situation. I believe people have canned recipes for IP Tables lying around.
That definitely sounds like a bug.
I did manage to find the article I was thinking about, but it was related to a different thing I was trying to remember. IEEE Spectrum, October 2005, "The VOIP Backlash".