Slashdot Mirror
Why Do We Have To Restart Routers?
jaypaulw writes "I've owned a WRT54G, some cheap D-Link home Wi-Fi/firewall/routers, and now an Apple Airport Extreme (100/10 ethernet ports). In the context of the discussion about the worst uses of Windows — installation in places where an embedded device is superior — I've gotten to wondering why it's necessary to reboot these devices so frequently, like every few days. It seems like routers, purpose-built with an embedded OS, should be the most stable devices on my network."
Fast, Stable, Cheap - pick two.
Most routers are cheap. (Apple's is overpriced-cheap; the point stands.) A bunch of them are free after rebates. Considering that, it's a wonder they keep running for more than 5 minutes. They come off the same assembly lines as those Norcent (who?) $15 DVD players.
You can buy reliable routers of course, from the C company, or the N company, or the J company, or a couple others. That's what corporations buy. What I wonder, though, is whether there's a middle ground: a "pro-sumer" router. Maybe somebody has got some suggestions.
If a client is able to cause a router to crash then there is something wrong with the router design.
And as you tread the halls of sanity, You feel so glad to be, Unable to go beyond. I have a message, From another time..
...the expectations of the user. Newsflash: when you buy cheap crap it is going to perform like cheap crap.
I want a new world. I think this one is broken.
That's what I was thinking. I have a linksys wrt-54g or whatever they are running ddwrt and I've probably has to reboot a handful of times in all the years I've had it.
If you have frequent power interruptions, aren't they rebooting your router frequently?
Beta sux! Join the Slashcott! http://hardware.slashdot.org/comments.pl?sid=4760465&cid=46173047
I've got 5 WRT54G's running DD-WRT out in the desert in enclosures. They routinely operate in temperatures in excess of 110 degrees, and usually have a bunch of users on them.
Is that degree C, F or K?
http://michaelsmith.id.au
crappy firmware. I flashed my WRT54G V4 with Tomato and haven't looked back. Also haven't had to reboot it in the past year or so that I've been using it, other than the occasional update. Tomato's developer obviously knows what he's doing: compared to the stock Linksys firmware he's lightyears ahead. And he's just one guy, you'd think a company with the resources of Linksys could do an even better job.
The higher the technology, the sharper that two-edged sword.
There's nothing about being "an embedded OS" that should make it any more or less stable.
Except for the fact that the end-user isn't messing it up? And that there are very few programs that are installed/will be installed? Really, most embedded OSes should be very stable because the cause for most OS crashes are A) Applications B) Drivers and C) user error. Because applications are not going to be installed and it should ship with very few to begin with, that takes that out of the picture. For B, the router shouldn't need any specialized drivers, or if they do, they should be minimal, and C) because the user only edits configurations that should be "safe" that isn't a problem.
Taxation is legalized theft, no more, no less.
Cheap "embedded" devices like routers and NAS-es routinely have extremely bad hardware. The competition apparently is so fierce that cutting corners of everything, from basic motherboard-like functionality to network and disk controllers is ubiquitous.
I'm occasionally doing hardware reviews for a local IT magazine and it's unbelievable what you can actually buy today as a bona-fide good equipment even from "brand name" companies. CPUs are usually ARM or AMD GEODE (You think VIA is slow? Think again. - Not to say there isn't a place for slow CPUs, only that this isn't it.), network controllers are cheap Realtek's and I don't know what they use for disk controllers (probably parts of the CPUs "companion" chipset) but it sucks.
I've seen "gigabit" network controllers on NASes that actually negotiate gigabit speed, although they are connected to buses and CPUs that break a sweat even at 100Mbit/s speeds. NASes that accept 4 drives cannot service reads on even one drive at more than 15 MB/s - introducing RAID (especially RAID 5) into this setup slows things to a crawl.
Practically all of these devices use Linux, because it's free (as in beer). They usually (I'd say 90%) don't acknowledge or obey the GPL.
It's a sort-of reverse "best scenario" for Open systems (and Open source). The manufacturers have a choice between something like this:
The first choice is represented by "truly" embedded devices like ordinary small, unmanaged Ethernet switches (with which I have suprisingly good experience), but apparently it's too expensive to scale it to "smart" devices that have to support many features so everyone opts for the second one. You can (and this is verified!) build yourself a small managed router or a NAS device like the ones sold at every el-cheapo computer shop with the same cheap generic components, and the resulting device will be just as sucky.
Creating a router or a NAS just like the above but with "proper" hardware (a Duron 800 MHz based system will be excellent) won't even cost you significantly more, but will deliver orders of magnitude better performance.
-- Sig down
That was my thought too when I saw this. If you have to reboot your router very often then you're doing it wrong.
We are heavy Internet users here at my home, especially since I work from home frequently and use VPN with the sending and recieving of large attachements, etc. I don't think I've ever *had* to reboot my WR54GT in the past year that I've had it.
I've noticed that ALL home routers at some point will require a power cycle, and not because they're bad, but because they all seem to occasionally lose their ability to provide DNS resolution. This isn't a problem on a LAN (like mine, obviously) which has a dedicated nameserver on the inside of the LAN, but for people who (like I once did) use their router as a nameserver.
" What luck for rulers that men do not think" - Adolf Hitler
Why DO you have to reboot your routers? Mine, including a WR54GT almost never require rebooting. Occasionally, after a power outage, it's necessary, but not very often. Maybe once or twice a year, and I live in Panama, where power interruptions come fairly frequently.
WTF? How did this end up +5 Informative?
The power interruptions are obviously regularly rebooting his equipment. Is it any wonder he doesn't need to reboot it himself?
Did it ever occur to you that you never had a problem *because* of the power failures doing the rebooting for you?
The most likely reason is usually overheating if you leave them on the carpet or somesuch.
I think the most likely reason is that he uses cheap crappy consumer routers.
last reboot was when I installed the new UPS and when I upgraded the kernel, respectively.
Stop Computers/Cars Analogies on S
OK. So other than bashing so called "cheap crappy consumer routers" and posting your uptime, what do you suggest? Not everyone has the resources to buy a $3000 Cisco router, nor does everyone have the time or knowledge to configure a Linux machine for routing.
DD-WRT on my WRT54GL, I've never had to reboot it for those issues. I even have a couple separate VLAN's set up, two DHCP pools on separate interfaces, etc. I've had uptimes of over 80 days before I tweaked something else on it that required me to reboot it.
It's not the hardware... it's the generic crap software that they run on.
My blog. Good stuff (when I remember to update it). Read it.
Stop buying shitty routers and you might have better experiences...I've used Netscreens (the old NS5/NS10s long since discontinued since Juniper bought them, but solid VPN/Firewall/Routers) for over 8 years - have easily had uptimes of over a year on some of them, and I've only ONCE needed to 'reboot' one (and that was after an especially bad lightning storm that nuked its UPS). It's not just the software/firmware/OS that leads to this (although it is a good part of it), but the actual physical components too.
I am like the previous poster. My Linksys WRT54G (Ver 3.0.3) has not been rebooted for a couple of years. I too didn't understand what the original poster was talking about, saying he reboots his router every couple of days. If you have a router that won't run more than a couple of days, I'd say you should get your money back. If enough people returned shoddy merchandise, the manufacturer would improve it (and might raise the price, but it would be worth it).
Instead of just bashing someone for using something cheap and not offering a possible solution.
the possible solution was implied (don't use cheap crappy consumer routers). also, to clarify, my routers arent $3k ciscoezzzz and don't run Linux.
For those of us that are at a minimum technically competent this is not an issue. [...] I was just thinking there are probably a lot of people that do not have the ability or want to tinker with firmware.
I can't understand why, what I like to call, "the grandma scenario" is constantly being brought up here (my grandma can't use Openbox / my grandma can't compile shit in the command line / my grandma can't update her fucking firmware). did the OP imply he is technologically retarded and can't use more than the router equivalent of a fucking toaster? this is Slashdot, not tech-tips-for-fucking-lonely-housewives. try to keep some decency, I mean look at your UID, for FSM's sake man!
why do cheap routers fail? well, here's a few tips: you paid shit for them, they're made of plastic and most of them are brightly coloured.
(apologies for going somewhat offtopic and using foul language)
Stop Computers/Cars Analogies on S
I don't know much about failure modes of Netgear routers, but your string of disk failures is alarming. Can you please provide some details?
Were these disks at home or in an office?
Were the failing disk in a server (with a lot of continual I/O) or a workstation with only intermittent I/O?
Thank you.
Fantasy: http://ferrisfantasy.blogspot.com/
You need to take better care of your stuff, you must be doing something wrong. Either that or you have the worst luck I've ever seen.
http://www.southparkzone.com/episodes/1206/Over-Logging.html
I fear the Y2038 bug
If you don't like the responses, maybe you should be browsing at +4 or +5?
There are many serious responses to this question that have already been posted. The good ones seem to cluster around the power issue. Which power supply sounds more reliable? A PC transformer is 3 pounds and $30-$70, a router transformer is 3 ounces and $3. Many slashdotters are software people, so the usual strategies of "blame the OS" or "blame the user" might be employed. The problem probably amounts to a hardware inadequacy.
I would never use a router without plugging it into the cheapest UPS I can find. Voltage fluctuates. The refrigerator compressor kicks on and the cheap Linksys transformer hiccups.
I've purchased many different versions of Linksys (some old, some new and crippled), Buffalo, Netgear, etc. They ALL go dead after a certain period of time if plugged directly into the wall. The WRT54G 1.1 transformer is physically larger than the 5.0 version. It probably burns more electricity, but it seems to be fairly reliable. I keep one upstairs as an ethernet bridge (no UPS and it has done fine).
Whether you're talking about routers or cable modems/DSL modems, the only way to ensure reliability is to fix the power supply. Are you running 12 gauge copper on a dedicated circuit to your router/switch/modem? Or just buy a UPS for $20-$30 and you can discontinue your daily or weekly reset of your cable modem. Your wireless access point will actually be on when you go to use it. And you can use your laptop or VoIP during a thunder storm.
No, I don't work for the battery company.
Four words: Switched Mode Power Supply. Switching supplies are extremely efficient and thus small. You've seen this already, there's one in your computer. That's how they can pack an 800+ watt supply in to something that small. You try and do that with a linear supply and it'll be massive.
Same deal for wall warts. Some companies still use linear supplies, but not many. It is to the point where SMPS is much cheaper to produce. They also have the advantage of being smaller, and running cooler.
You can see the technology at work in amplifiers. Have a look at Rotel's site sometime. Specifically, compare their old RB-1080 vs their new RB-1092. The RB-1080 is about twice the vertical size as the RB-1092 to house some large components (the transformer and the heat sinks mainly). However it outputs only 200 watts per channel where as the 1092 does 500. How? Well the 1080 is an older style amp. It uses a linear power supply, and class AB output. Thus it takes a lot of space and generates a lot of heat to do what it does. Not very efficient. The 1092 is a class D amp. It has a switching power supply and switching outputs. As such it is very efficient and produces little heat.
Now none of that is to say that Dlink isn't including poorly made SMPSes, however the fact that they are so small has nothing to do with their quality. You don't need big if you have efficient.