Take the 2-minute tour ×
Server Fault is a question and answer site for professional system and network administrators. It's 100% free, no registration required.

I've been curious about this for awhile now. I don't know if I'm seeing a real pattern or not, but having worked with many home office/small office wireless routers for a few years now, I've noticed that the heavier the traffic is in the environment, the less reliable they become over time.

Desktops become clogged with malware and innumerable user installed programs, but a router just sits there "untouched" - pretty much the same as the day that you pulled it out of the box - other than occasional firmware updates.

I'm not talking about total failure, but the frequency that they become unresponsive. A quick reboot and users can connect again, but over time those reboots happen 3-4 times/wk rather than once a month or once every 6 months.

I guess typically over time bandwidth demands also go up - more users connecting or downloading/streaming more things, but in a couple environments where usage rates stay relatively constant, it still seems to happen.

I would think that the hardware would basically either work or not rather than a general decline. What components, hardware or software, might be causing this or is it a made up pattern in my head?

Edit: Not to make this question more complicated, but sometimes it's just the wireless access that stops. The wired computers continues to work just fine. Hopefully, some others can confirm that they've seen this, too.

share|improve this question
3  
I've had the EXACT same experience over the years across several routers. I'd love to know if someone knows the answer to this. –  Electrons_Ahoy Jun 5 '09 at 1:44
    
Regarding the edit - that IS the exact problem I've been having. Maybe it just the store - Keith, you don't shop at the Sacramento Fry's Electronics, do you? :) –  Electrons_Ahoy Jun 5 '09 at 6:18
    
There are a lot of interesting answers/comments here. Thanks everyone for the discussion. After reading, I think that probably the most contributing factors are: heat degrading the hardware over time, increased interference from the environment (other devices and power issues), and sometimes partial hardware failure (transmitter degrades faster). Hope that's a good summary. –  Keith Bentrup Jun 9 '09 at 5:58
    
We're talking about a $40 device that cost maybe $20 to build, it gets run 24/7 for a year or two, chances are the components aren't spec'ed for that kind of time period, plus what everyone said about heat/power issues. Also consider these things tend to be run on really cheap software (typically some stripped linux image) so who knows what kind of internal problems it is having software-wise. Personally I just buy a new one every year or two and pretty much treat all that stuff like the disposable junk it is. –  Kurt Jul 1 '09 at 5:43
    
I must be lucky - I bought 2 Bufallo linkstation WAPs in 200..2? maybe and then I sold one to my boss in 2005. Both are still going, mine having an uptime of ~3 years under everyday use! OTOH, my parents are presently going through wireless routers like they are going out of fashion.. maybe we need to add "dumb luck" to the answers? –  Tom Newton Jul 1 '09 at 8:27

13 Answers 13

up vote 11 down vote accepted

In my experience, it's been heat. There's not a lot of airflow in these devices (no fans) and they're usually tucked up in closets or somewhere with no airflow. My last wireless router was screwed into a wall high in a closet. It lasted a year, and the case was always warm. When I replaced it (with the same model of linksys), I put a fan up there with it and the current one has lasted three years.

share|improve this answer
    
I wouldn't be surprised if heat had something to do with it. Certainly more traffic = more processing = more heat, but what would it be doing inside that heat would cause it fail intermittently? and why would a reboot take care of it for a couple days? It's not like a quick power cycle would give it time to cool, right? –  Keith Bentrup Jun 5 '09 at 2:25
    
I Am Not An Electrical Engineer, but ... Intermittent failures can cause firmware to hang. Heat, or rather the expansion and contraction of metal bits from exposure to heat, can cause intermittent failures or unexpected states in the integrated circuits that make up the controller. It's not so much that you're letting it cool when you power cycle, you're just clearing anything stateful and reloading the firmware from ROM. –  Karl Katzke Jun 5 '09 at 2:50
    
So the hardware failure caused by heat causes software failure. Then we reset the software, and why doesn't the still hot hardware cause the software to fail again almost immediately? I'm missing something. I think heat plays some part in it. It makes sense to me that the more that these expansions and contractions occur, the more they occur - a self-feeding cycle - due to the properties of the materials used. Maybe they become too malleable or brittle. Where's a material scientist when you need one? :P –  Keith Bentrup Jun 5 '09 at 7:01
    
Software failure in the case of a lockup usually isn't a case of an immediate failure like a short. Think more of a traffic gridlock in a downtown. First you get an accident at first and Main, then one at 3rd and Front, then one on the on-ramp to the freeway and everyone's stuck. But three wrecks don't always happen... in the morning, you only have one wreck and things keep moving. Then in the evening you have six wrecks and no one gets home till 8pm. –  Karl Katzke Jun 5 '09 at 11:47

Many of these small devices have a hard time with power sags and brownouts. If the power goes completely out (and stays off for a few seconds), then comes back on, things are OK, but other power problems cause lockups or other strange problems (returns pings, but won't route).

When we moved into our house 8 years ago, our power was rock solid for several years and my Linksys WRT54GL never missed a beat. During the last few years, there has been a lot of nearby construction and power sags and outages have become relatively common. If the lights flicker, often my computer will keep going, there is a 50% chance that the coffee maker's clock will reset, but there is about a 90% chance that the WRT54GL will lockup. I assume a UPS would fix the problem, but just I haven't gotten around to trying it. At some point, the wall transformer failed. It seems that the WRT54GL has been better since it was replaced.

Also: Many of these devices run Linux (mine runs openwrt). The more traffic you push through a typical Linux device, the more likely you are to run out of some type of kernel resource (slots to keep track of TCP or NAT sessions, process handles, etc.). Rebooting the device will reset the tables. We had this type of problem with Linux based firewalls in our data center. I never was able to track down what kind of resource limit was being hit, but adding RAM helped and finally a substantial hardware upgrade (running the same software) has fixed the problem completely.

share|improve this answer
2  
UPS's work wonders for connectivity. I've been here 3 years with my wireless router and I haven't had an issue because it's been plugged into a UPS (with lots of power fluctuations). –  JFV Jun 5 '09 at 3:58
    
Man, I had never though of that, but now that you bring it up I'd be willing to bet that the wiring in my 120 year old house is the problem. –  Electrons_Ahoy Jun 5 '09 at 6:15

Two reasons heat and age.

  1. Inexpensive home routers have very poor heat dissipation, causing lockups and poor transfer rates.
  2. Transmitters wear out over time (probably due to item #1)

Every WiFi router I've owned has lost the transmitter eventually (usually long before the wired connections failed.) You can extend the life of the router by mounting it in a cool dry place. The one in my (nice clean) basement has lasted three years and is still going strong. On top of your (dust covered) TV it will die in 1.

share|improve this answer
    
One thing you can do is to throttle down your bittorrent upload speed to about 80% of max. This keeps the router much cooler than maxing the connection and ensures overhead for web surfing too. –  Chris Nava Jun 5 '09 at 4:08
1  
Not to be picky, but "age" isn't really an answer to the question why does X happen over time. It's like saying b/c time has passed. The transmitter part narrows it down to a certain part of the router, but what is it about the transmitter that would be less reliable? –  Keith Bentrup Jun 5 '09 at 6:36
    
Not being an electronic engineer, I can only speculate that its caused by prolonged heat damage. I do know that home routers are much more reliable when kept cool and dry. –  Chris Nava Jun 13 '09 at 3:22

One thing that comes to mind is that maybe it's not the device, but that the environment has become more noisy. More microwaves/phones/cell towers, and more other wireless networks in closer proximity may be giving your router fits.

share|improve this answer
    
Hmmm .... interesting possibility. –  Keith Bentrup Jun 5 '09 at 2:06
    
Had an experience like this with one of my first 802.11b access points in a production environment. The boss's desk was in line-of-sight from the antenna. Between the boss's desk and the antenna was the kitchenette with the microwave. Every day around noon, the boss's laptop would drop it's connection... it took us an embarrassingly long time to figure out. –  Karl Katzke Jun 5 '09 at 3:19
    
I pretty much always lose wireless internet in the living room when my Wii-motes get turned on. Access is fine everywhere else, but those wii-signals are just too strong for my poor router. –  Electrons_Ahoy Jun 5 '09 at 6:17
  1. Logs. Stop guessing and check the logs of your devices. Usually 192.168.100.1(Modem) 192.168.1.1 (router), and admin/admin will get you into 80% of the devices. But google the model number for more info. It will tell you a lot of what is actually causing the problem.

  2. Troubleshoot before resolving. Many times its human nature to start trying to solve a problem before you understand what the problem is. When you connection bogs down, what is bogging down? Slow load time(Pass through)? High ping? Low wireless signal? Delay response(nothing,nothing..boom, 100% load), etc. And don't forget you can have more then one problem. Investigate, Isolate, Eliminate.

  3. Don't rule out software. You point out hardware should be stable. Maybe it is. Could you be getting some interference and thus another wireless network draws the eye of your client's roam? Could heavy computer load be messing with the encryption algorithm? Most important, what kind of AV you got? And even more important, don't use the router's propertery wireless client. Always, always, always use OS client. Big reason why mac networks seem so much more stable. Unless where talking work with VPN or such, then your stuck.

  4. Check what settings you have on your router? QOS? Disable legacy wireless settings (IE disable B and do G only), Is it secure? Whats your NAT type? UPNP? Ask a complicated question and receive a complicated answer. Good luck my friend.

share|improve this answer
    
I've found too often that when routers lock up, one can't get to the admin page. Perhaps next time run a syslog server on your PC or another machine, have your router send its logs there and hope the smoking gun is there someplace. –  dmoisan Jun 8 '09 at 14:20

At home, it is mostly the neighborhood wifi in use, Microwaves, and things like cordless phones, wireless home theatre speakers etc. The best way to assess the best wifi channel to use, is to use something like netstumbler (or whatever is the latest). Or trying to use Wireless-N and check if the issue still persists.

share|improve this answer

Entropy :(

share|improve this answer
    
Unique answer. :) By this, do you mean that over time there are more ways (states) that it can fail than continues to work, so failure becomes more probable? Or did you mean something else? –  Keith Bentrup Jun 9 '09 at 5:52
    
That's precisely what I meant :) Nice someone picked up on it –  Chopper3 Jun 9 '09 at 6:13

I would suspect that like a lot of hardware , the electrolytic capacitors have dried out. They are filled with wet paste that makes the electrolyte: this makes the capacitor have such high capacity. As they dry out they loose capacity ( and capacitance). After replacing the caps it will probably run fine. (Probably need to replace the ones in the power pack as well.)

There are websites dedicated to the problem of "bad caps", which are cheap (generally) Chinese sourced parts that fail early (1 year). The Bad Caps forum

regardless of manufacturer, running hot they will last only 3-5 years.

When making extended temperature range kit (military/aerospace) the solution is to use tantalum capacitors instead. Tantalum capacitors cost a lot more. They also have a scary failure mode where they explode after a while generally weeks , when installed backwards. AKA Tantalum time-bombs.

share|improve this answer

I've got routers / APs that have been in place in Customer sites for 4+ years with no problems (some "enterprise class", and others that are just Linksys toys). The AP that I'm using right now is a 2003 vintage Linksys WRT54G, and it works as well as the day it was new.

I suppose some hardware components in some devices could be borderline on tolerances, and years of use could cause them to degrade. Being in a harsh environment might do it, too. Generally, though, electronics either suffer "infant mortality" or run forever. (I could see the little wall-wart power bricks on some units going out of spec as being a potential cause, too. Some of the lower-end units come with pretty crappy power adapters, and I have replaced a few of those over the years with "generic" universal adapters from Rat Shack.)

I guess, ultimately, I'm saying "I dunno." I've got no good answer for you. >smile< Wonder why I posted, eh?

share|improve this answer
1  
In environments where there's rarely any heavy traffic, I don't often see this. For example, a home office with a single wireless laptop. Where I do see it happen is when 1 off the shelf router is shared between 10-20 coworkers or even a home with kids doing a lot of peer to peer sharing. –  Keith Bentrup Jun 5 '09 at 2:12

I've had the same experience as you - I've been through numerous WAP/routers over the years. My current Linksys WRT54G is several years old, and now it needs to be rebooted about twice a week. It didn't used to be that way.

I expect it's a matter of dust building up inside and maybe tin whiskers on the components.

share|improve this answer
    
But I would expect excessive dust or tin whiskers to cause shorts or complete failures, not intermittent behavior. Do you know of other examples where dust/tin whiskers cause a similar pattern? –  Keith Bentrup Jun 5 '09 at 2:06
    
I used to work in a factory, and the lathe/cutter controlling computer would play up wuite often - a reboot would generally cure the problem, maybe for a day or so. In the end we discovered that blasting with an air duster stopped it from happening - it ended up being one of our weekly maintainance things. Aluminium shavings ... –  Tubs Jun 5 '09 at 13:05

I'd really reconsider the use of wireless for a network where bandwidth is heavily demanded. Specially if you don't have the required policies (features or capabilities) in the Access Point(s) or controller to prioritize and/or shape said traffic. Sure there are cases where access points (or any kind of device) just gives out due to temperatures, static, bad signals, northern lights, etc... but talking about wireless, I think it's more common for this kind of network to start becoming less reliable due to heavy loads. I'd make sure you have the correct specifications (as in enough equipment) to support this, if wireless communication is a requirement for business-critical traffic in your environment.

share|improve this answer

This at least partially comes down to the fact that electronic components do degrade over time, capacitors being the worst. Unfortunately, this affects a number of things but perhaps the biggest problem for our purposes is frequency drift. Yeah, I know it's supposed to be compensated for but that compensation is far from perfect, as it also relies on components which degrade.

The reason we see this affect wireless more than cabled connections is simply because wireless is very sensitive to frequency changes. Couple this with the causes noted by others and it's a wonder that wireless devices don't suffer even more than they do.

share|improve this answer

I've noticed my four year old Linksys blue box works fine until I start torrenting. I've seen explanations that this has to do with the memory on the device, it has to start tracking lots and lots of hosts in the state table and it goes nuts. It doesn't need a reboot, but all connections bog down until I turn bit torrent off.

If I tell my torrent client to max out at two torrents, however, things are fine. It wasn't a speed issue, it was a "number of connections" issue.

share|improve this answer

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.