10

My question is a bit general but I assume there has to be a good, specific answer to my problem.

I, like many, am beginning to wonder if my ISP is throttling my home internet connection. I'm not so quick to judge because throttling accusations are often unwarranted or unsupported and often seem a bit conspiratorial in nature. I'm not yet ready to make that conclusion.

In my case, my wife and I get our internet through our cable provider, and we have the top-tier, home-based offering for our area. We have found that very consistently, our ability to surf the web between 12:30PM and 2:00PM completely disappears. I understand that this is typically lunch time but, from what I can gather, this isn't considered peak time. Furthermore, we often stream Netflix in the afternoon during what is certainly peak-usage times (8:00PM-10:00PM) and we never have any trouble.

If this is an issue with network load, we should have trouble in the evening, not in the early afternoon.

What could be causing this mid-day bottlenecking? This happens every single day, and it happens on every device (we have over 7 phones, tablets, PCs, etc. and they all have the same problem.) Could this realistically be an issue with ISP throttling? Could it be something else and are there steps that I can take to personally troubleshoot or diagnose the problem?

Regardless, this is extremely annoying and I'd like to do a bit of personal troubleshooting before I call up the ISP and ask them for a bit of customer service. Our ISP's customer service is horrible and they've never been of much help. If I am going to try and confront them on this issue, I would like to know as much as I possibly can about the nature of the problem.

Any tips, strategies or pointers from those who are well informed would be much appreciated.

animuson
  • 273
  • 1
  • 3
  • 16
RLH
  • 4,185
  • 16
  • 42
  • 55
  • Any updates? Have you solved the problem? Can you tell us what were the results of ICSI Netalyzr and trying Ethernet instead of Wi-Fi at lunchtime? Curious minds would like to know.... :) – allquixotic Aug 15 '12 at 14:09
  • allquixotic: Sorry for no response. I haven't abandoned this question but to be able to test out a few things, I need to be home when the problems occur. I work away from the home during the day, so I probably won't be able to update this post/mark an answer until this weekend. – RLH Aug 15 '12 at 14:31

4 Answers4

8

How to diagnose ISP throttling

ICSI Netalyzr is the gold standard for analyzing what nasty things your ISP may be doing to your internet connection. But also keep in mind that many problems are related to your "endpoint devices" (DSL/cable modem, etc).

Also check if you have any background programs that do a daily upload of a lot of data, like Steam syncing to the cloud (although that's not daily), Dropbox, or something like that. Saturating your upstream can obliterate your downstream, which would do exactly what you described. This is because of the TCP ACK stalling problem, which is present in every single endpoint device with an asymmetric link (less upstream than downstream or vice versa) and which doesn't have Controlled Delay Active Queue Management implemented (CODEL is a very new technology that just appeared this year, so consumer devices are practically guaranteed not to have it).

You might also have issues if someone in your neighborhood routinely does something at the same time every day, like if they vacuum, or cut the lawn, or turn on a sprinkler, or try simultaneously downloading 100 torrents, or something like that.

You'd be amazed how little it takes to interfere with a cable connection. And 99% chance it's not intentional.

TL;DR:

  • ICSI Netalyzr is a great tool for diagnosing throughput or latency issues (run it during "normal times" and during the "slow times" also)
  • Saturating your upstream will bring your downstream to its knees, because your cable modem doesn't have CODEL (trust me, it doesn't) -- buffer bloat and TCP ACKs expiring, etc.
  • People can mess with the line just by doing ordinary activities, and it affects you too, not only them, even if the cable is buried in the ground
  • Water can disrupt a cable that has a nick or scratch in it, i.e. if the cable housing isn't totally intact
  • Cable modems themselves can be weird and decide to stop working, so if you have one older than ~2010, it's time to upgrade to the latest model

Edit/update: Are you also on Wi-Fi? :) Wi-Fi is bad; if you are using it, stop using it! It's 100000 times easier to disrupt Wi-Fi (intentionally or not) than it is to disrupt a cable in the ground. Microwaves, cordless phones, vacuum cleaners, and almost any other electronic appliance or device can carelessly spew electromagnetic pollution on the unlicensed Wi-Fi bands, either because that's part of its operation, or because the designers were lazy and didn't measure/control EM emissions from motors or sensors within the appliance. The unlicensed bands have an incredible amount of "noise" -- some of it extremely "loud" -- being flung all over it at any particular time, from countless sources. So stop using Wi-Fi and your problems will probably go away.

allquixotic
  • 34,215
  • 7
  • 113
  • 145
  • RE: WiFi comments. Yes, we use WiFi but I don't understand what would cause this problem to occur for only 1 hour of the day, always at the specified time, every single day. I will try to directly plug into the router, though and see if that helps. I should have tried that first but I didn't think about it since our WiFi has worked great, otherwise. Of course, the ideal (for me) would be to give up WiFi, but that would involve getting our house wired, plus, some of these devices don't have a way to directly wire into them. (iPhone, iPads, Apple TV... yes, we are an Apple house.) – RLH Aug 13 '12 at 19:28
  • Also, thanks for the ICSI Netalyer link. I'll be sure to give that a run this evening, and tomorrow. – RLH Aug 13 '12 at 19:29
  • If only you had Geordi's visor from Star Trek: TNG, you could visually detect all the ridiculous EM pollution we have in our daily environment. It could be something as weird as a malfunctioning "wall wart" (one of those little black boxes you plug in to convert AC to DC for some appliance) -- when those go up, they can emit some strange EM. Every vacuum cleaner I've ever owned spams 2.4 GHz. And every microwave, and every cordless phone. The point is, the Wi-Fi spectrum is **unlicensed**, so spamming on it for any reason is **legal**, hence appliance makers don't care if theirs does it. – allquixotic Aug 13 '12 at 19:45
  • And keep in mind that the source of the interference doesn't have to be within your house. Unless you live in an **extremely** rural environment, your neighbors could be doing it to you. Maybe the neighbors have a maid that comes in and vacuums the entire house from 12:30 - 1pm every day? Something silly like that could disrupt your WiFi. – allquixotic Aug 13 '12 at 19:49
  • That would be odd, seeing that his happens every day, even on the weekends. Still, I get your point but, if this was EM interference wouldn't the quality of our connection degrade? i.e., wouldn't our signal strength drop from, say 4-5 bars to 1-2? This doesn't seem to happen. – RLH Aug 13 '12 at 19:52
  • That kind of depends. The signal strength meter shows you an average strength over a period of time. If the time period is short enough it bounces around like a troubled child on crack (but doesn't help you much, because there is error checking in WiFi nets). If it doesn't react enough you never see major failures happen. Point is, it's a consideration, but no where near the BS level of "100000 times." WiFi is different, not "bad." – Everett Aug 13 '12 at 19:59
  • @allquixotic - Can I ask where you got your data from regarding this claim, "It's 100000 times easier to disrupt Wi-Fi (intentionally or not) than it is to disrupt a cable in the ground." Next, you're not talking about putting a cable in the ground, he'd likely have it in a wall (if he wired his home). RFI interferes with Cat5, 5e, and 6 all the time (I've diagnosed it before). – Everett Aug 13 '12 at 20:05
  • You basically need a clean room / controlled environment to get any sort of meaningful reliability out of wifi; that's why I say it is bad. It's not so much the protocol, but the fact that it's unlicensed, and so many things either use it to transmit information or pollute it with junk noise, that chasing down a transmitter of interference in a typical urban environment is nigh impossible. It's just not worth the effort. As for your WiFi-only devices; I sympathize, but let the buyer beware... I would personally never buy anything that isn't either wired ethernet or CDMA/EvDO/LTE. – allquixotic Aug 13 '12 at 20:08
  • @Everett - "100000 times easier" isn't a scientific measurement, just a comparison. Cat6 cable that's actually manufactured to spec (or above spec) is very unlikely to be interfered with unless it comes very close to or touches certain kinds of other cable that are themselves poorly shielded or unshielded. The point being, you have direct control over what your cables in your house are or aren't touching, but you have NO control over what EM enters your house from the outside, unless you install an all-encompassing Faraday cage. And it's hard to replace devices emitting 2.4GHz pollution. – allquixotic Aug 13 '12 at 20:12
  • I can just add, it's worth downloading the Command Line client, as it may be wise for other reasons not to use the Java web plug-in. – RipperDoc Feb 26 '13 at 13:44
3

It is unlikely that this is throttling. Throttling is typically based on an activity, not a time of the day. If you were running a torrent client every day during those times, that might cause some interest by your ISP. However, most ISP's (including one's I've worked for) throttle the offending port, not the offending IP (web browsing would still work, but torrents would be slow).

Now, how to diagnose.

The truth is, you can do little to diagnose outside your network. This is intentionally put in place by your ISP, because things you do to diagnose outside your network, are the same things people do to learn about a network before attacking it.

Have you checked to see what your upstream bandwidth looks like during those hours? Most ISP's are happy to provide a document that shows your usage. You may as well ask for downstream usage too.

You may have something on your network that decides it's time to "phone" home everyday at the same time. This could be an app you installed, or one someone else installed.

If you discover high usage in either direction, ask for a report that shows what ports the utilization is on. This way you can cross reference port usage and determine what software is using bandwidth.

Do this with the attitude of, "Dear ISP, I may be inadvertently causing you a problem, and would like the following information to determine if that is the case..." If you decrease your utilization, that decreases the amount of use on their network. They have no reason not to help you. You may even have to remind them of that.

Let me know if you have any other questions.

Everett
  • 5,912
  • 1
  • 22
  • 33
  • I'll take a look at the one PC that stays connected at our home during the day. I doubt that's the problem, though, but you never know for sure. – RLH Aug 13 '12 at 19:37
  • Have you ruled out someone connecting to and using your WiFi connection every day at lunch? Maybe some high school kid wants to D/L porn, and has set a Chron job. He uses your network so mom and dad don't catch him... – Everett Aug 13 '12 at 20:07
  • Doubtful, but worth checking. The device is WPA2 secure with a long, cryptic passcode. – RLH Aug 13 '12 at 20:25
  • Just checking all the silly gotchas. – Everett Aug 13 '12 at 20:27
3

Measurement Lab (M-Lab)

is good overall resource if one is concerned about ISP traffic management policies, and how it may be affecting performance

It's a joint effort by industry (Google is a headline sponsor) and Academia. They have a number of tools available

Of the listed tools, I tested ShaperProbe (from GA Tech) and found it was useful in identifying shaping levels occurring on my home Comcast connection.

If you're really passionate about this, Project Bismark allows you to run a mini performance lab from home. It uses an OpenWrt flashed Netgear router to provide continual monitoring and sharing of performance data.

cavalcade
  • 157
  • 8
1

The most likely cause for your lunchtime slow speed is a poor configuration of the network from your ISP. Unfortunately, there's nothing you can do (apart from organizing a class action :)...)

Most ISPs use what's called "overselling" - they sell "up to 100Mbps" knowing upfront that if all their customers in one area (i.e. connected to one particular network gateway router) browse or download (read: stream Netflix) at the same time, everyone's speed will be slow due to the bandwidth management setup. As this happens rarely, they get away with it, make a good profit and you (the customer) see it as an annoyance and live with it.

I assume you're only one family of many today having the same habits during lunch time - streaming HULU and Netflix is a cultural phenomenon. Technology, like any other business area, move ahead driven by consumer demand - ISPs are not the only ones caught unprepared by the current bandwidth thirst :)

I am just guessing here, but this is most likely the cause of your problem. In order to solve it, you can only look for a new provider - one that offers fibre or ADSL or - if available - a wireless Internet provider. These are usually better at managing increased demand - provided they have the infrastructure in place.

Just as a note: the more popular an ISP is, the more prone to "oversell" - it is a big job to overhaul a network infrastructure once it is in place. But it's also silly to refuse to connect more customers on it at the same time :))

Hope this helps a bit

Cris
  • 11
  • 1