More Recent Comments

Monday, March 26, 2007

Internet Connection Speeds

 
Here are the results of a test for internet connection speed from InternetFrog.com. This is the speed I get at the university.

Now, here's some questions for all you technical experts out there. The download speed varies from a low of 2 Mbps to a high of close to 9Mbps. Why? Does anyone have faster connections on a regular basis?

The upload speed varies from a low of about 150 Kbps to a high of 1.7 Mbps. Why? And why is the upload speed so much slower than the download speed? Does that depend on the speed of my processor?



[Hat Tip: Kevin Black]

14 comments :

Unknown said...

I'm not an expert, but download and upload speeds are genrally capped by ISPs.

If you have an ADSL connection, the "A" stands for asymetric--upload speeds are slower than download.

Your speeds are phenomenal--about five times faster than my usual speeds.

Ruth Bridges
Vancouver Island, BC

Eamon Knight said...

I doubt a major university runs on ADSL, so the UL/DL asymmetry puzzles me. I'm not really an IT expert, but I would have thought that Larry's computer is connected to a building Ethernet, which connects to a campus WAN running IP (possibly over ATM), which at some point routes into big bit-pipe like a T3 or better connected to the Internet.

But why ask us? I'm sure DIG knows how all the t.o traffic gets in and out of your lab ;-).

Anonymous said...

I can't check my own speeds, due to some sort of missing plugin, but anyway,

NO, your CPU is not limiting your upload speed. A modern CPU should be able to handle at least a couple hundred Mbps. We use a lot of gigabit connectivity, and we were seeing 20-30% usage, and that was with CPUs of 3 or 4 years ago.

Anonymous said...

* and, it is more likely that we were limited by hard drive speed, not CPU.

J said...

Having worked for an ISP for a year and a half in the past, and knowing the on campus demographics of a university, I can't tell you almost for certain that the UL/DL asymmetry is due to file sharing. If a student has their computer on and conneted to a file sharing network and still has all the files they've downloaded in the past shared out, there are probably far more people connecting to them and downloading (upstream relative to the university) than the student is making up for with his/her downloads. They don't even have to be downloading anything for someone else to be downloading from them. That basically means that all the people downloading files from students on the university are chewing up the upstream bandwidth.

It wouldn't surprise me at all if they didn't have any of it blocked. The ISP I worked for in 2005 didn't have any file sharing blocked, and our customers complained for months when we started shaping.

Anonymous said...

Not a super techy guy, so if someone who is comes along... but until then:

The upload and download limits, AFAIK, are set by the ISP and are typically asymmetrical (within those limits your mileage may vary, in fact will vary). There's no technical reason for this; it's just cheapness and lately the desire to have different levels of service at different levels of payment. The assumption is that you want to download fast far more than you want to upload fast, and is usually correct. But most folks wouldn't complain anyway simply because when speed is mentioned by ISPs you'll notice it's usually DL speed and no mention of UL speeds.

Larry's very fast connection may be because the university there has a good setup, and perhaps because there happens to be a pretty direct route to that particular speedtest (for instance, I used to get fantastic DL speeds from Tucows' site in Vancouver, probably because it had a pretty direct, fast set of connections to Shaw out here (Vancouver island).

You may have noticed in reading around the web that some countries have much better broadband speeds than others (I belive I've seen South Korea mentioned re this) laregly because folks demand it and ISPs then do it. It's somewhat like the situation in photography that I used to see mentioned -- that European consumer photo labs typically did a much better job, apparently because their customers would complain if it wasn't really good while the customers in North America would tend to simply accept a crappier end product.

Trinifar said...

The download speed varies from a low of 2 Mbps to a high of close to 9Mbps. Why? Does anyone have faster connections on a regular basis?

All I can add is that I sometimes experience really, really fast download speeds that lasts for 30 minutes to an hour (of course I'm not at your university so it's apples and oranges). On those rare occassions when the speed goes way above my normal 3 to 4 Mbps, using the Internet is a truely different experience. I wonder if it's just a rare confluence of random factors (e.g. some major users of bandwidth have paused for maintenance or a holiday) or if my cable company has taken off the governors for a while to do some tweaking.

Maybe it's just sunspots.

Larry Moran said...

trinifar says,

Maybe it's just sunspots.

Have you checked the phases of the moon?

Maybe God does it.

Anonymous said...

You are really just testing the bandwidth of your connection. Back in the days when everyone was on dial up your bandwidth was tiny and would create a bottleneck. Now everyone one is on broadband and your isp can make lots of money by sell bandwidth and equating it to speed. To really test the speed of your connection you need to check your "ping" time. Even better do a trace route and see how fast and how you connect to individual servers and sites. From the command line >tracert sandwalk.blogspot.com.

Tim said...

You may also want to try some other connection speed testing toolds. I did a little writeup on the free one offered by speakeasy.net.

http://timarcher.com/?q=node/28

Anonymous said...

We all know that broadband suppliers advertise their speeds as "up to"
a certain level. But how fast is your actual connection? Now you can
find out with our broadbandspeed test, use the
checker below to find out. To get accurate results make sure that you
are not using your internet connection for anything else while the test
runs (it only takes a few seconds)

chas said...

I am not an expert but I have done a fair bit if research into Internet speeds and various types of services. Here are my opinions:
1. There are two main types of very high speed services here in Canada: ADSL (Bell) and Cable (Rogers). There are other service providers (ISPs) but these are the main two.
2. ADSL - is the telephone line into your home and is like spokes on a wheel with the source of the service at the hub. The concept is no impact to one user if there are other users on the same hub. This speed is definitely slower (upload and download) than Cable.
3. Cable - is via the TV coax cable into your home and is shared by other users in your area...impacted by numbers of users and amount of downloading.
4. ISPs know there is much less requirement for uploading capability and therefore restrict this to about 1/4 to 1/10 of there download service.
5. The term high-speed is a misnomer these days, as it could mean as low as 150 Kbps or as high as 10 Gbps (offered by the Cable companies at a premium).
6. I have Rogers Cable Extreme which can provide a download speed of up to 10,000 Kbps (or 10 Gbps). 7. I have measured my speed (via the speedtest facility Larry used) as high as 8000 Kbps. However, my upload can only reach about 900 Kbps).
8. Having said all this, now (see the Rogers article in today's Star), it is quite apparent the major ISPs also appear to selectively reduce your speed depending on how much downloading you are doing. This is not how the service was advertised ... but they want to change the rules and have us pay more. Sounds business like, right?
9. Having said all that, I am a BitTorrent user and have noticed Rogers reducing (dramatically) my download speed...without advising me. This does annoy me somewhat.
They do say "...95 percent of its subscribers were at risk of seeing their service deteriorate because of the surfing habits of the other 5 percent." A great way to twist the facts!
Firstly, if Internet users are seeing service deterioration there has been no mention of a complaints of slow service. Secondly, this method is a good way of justifying a raise in price and blaming others for their poor hardware planning.
10. The real problem is, like a lot of x-Bell users, I moved to get away from Bell's terrible customer support and extremely poor ADSL service (did you know they only guarantee half their advertised ADSL speed?) ... only to now find Rogers is deciding they don't need to be as competitive anymore.
11. Time for a new major ISP to emerge.

Teja said...

hai,
I performed my internet connection speed by using the http://www.ip-details.com/internet-speed-test/.My speed test results are,
7.02 Mbps when downloading,
2.86 Mbps when uploading..

chas said...

An update to my May 30th comment. Rogers has changed philosophy and now charges for downloads in excess of a 60GB or 80GB limit for residential users. Their speed is now consistently close to advertised. I noted these speeds this morning (http://www.geocities.com/web_mgr/graphics/speed.jpg). For 10Gb/s speed and 60GB data transfer, I pay $35/mo (for 6 months) and then $45/mo thereafter. I do download a lot of movies (not illegal in Canada, yet)...