Most Home Broadband Users Don't Know Their Speed

[sechs]

I guess it working correctly is a much higher concern....

http://news.yahoo.com/s/pcworld/20100601/tc_pcworld/mosthomebroadbandusersdontknowtheirspeed

 

[ddrueding]

Most users shouldn't have to, really. Once it is fast enough to stream your video in real time, that is all you need at home. That is why broadband in this country isn't getting much faster; normal users wouldn't notice if it was.

Now that Youtube has started offering 1080P, users might notice. Once Netflix starts offering really high resolution video, it will be more significant.

 

[LunarMist]

Aren't there numerous sites that test download/upload speed? A quick check of the Google should find useful results. :drnk:

 

[BingBangBop]

Actually, I find with My Comcast that the DL/UL testers are not doing a good job. They tend to be single threaded but Comcast seems to be BW limiting a single stream. I can triple my tested and stated BW by simply running three independent threads.

I found out about this by simply uploading a single very large file and it capped out very close to my supposed upload BW. Well, this file was over a GB and it was going to take 3+ hours and I had 7 more equivalent files to upload. Well, I didn't want to wait for it to finish and start another when it was done. I wanted to go to bed instead. So I started another, and to my surprise, watched as my transfer rate doubled. So I started a third upload and watched as my transfer rate tripled over what my account says my maximum BW is supposed to be.

I started 4 more and the BW stayed at 3x what was supposed to be my maximum. I was very happy now that I could go to bed and when it got up it would all be done. When I got up everything was finished. What was supposed to take 21 hours of uploading took less than 7. I can not attribute this to any form of caching or short-term BW boost technology in that these were massive files done over a large block of time.

I have several times checked and the tested BW using ndt servers is still limited to 1x (768Mbps UL and 12Mbps DL tested today to get ~699Mbps and ~10.99Mbps) both for UL's and DL's. I have also tested it again by uploading and downloading files and I can still consistently maintanin 3x those speeds by just doing things in parallel both for uploads and downloads - Nice Comcast, Good Comcast. Do note that this is during "off hours". Doing the same multiple massive uploads during daytime simply results in all the upload FPT sessions being unceremoniously terminated within 10 minutes -- Bad Comcast, Evil Comcast but really rather understandable considering the UL limitations of cable (It is activities like this that really interferes with other Comcast users broadband experience). Daytime HTTP Download sessions don't get terminated though but they don't go at 3x speed either.

 

[MaxBurn]

Reliability is far more important to me 3360 Kbits/sec down by 864 Kbits/sec up isn't terribly fast but the thing almost always works (lightning storms the exception) and more important to me it is always that speed. I regularly max it out for hours at a time without interruption, up or down.

 

[Mercutio]

Youtube does offer 1080p streams, ddrueding.

 

[LunarMist]

Reliability is far more important to me 3360 Kbits/sec down by 864 Kbits/sec up isn't terribly fast but the thing almost always works (lightning storms the exception) and more important to me it is always that speed. I regularly max it out for hours at a time without interruption, up or down.

That is my philosophy as well. Cable around here is unreliable. Sometimes it is out for hours, which is intolerable.

 

[Mercutio]

I can play video files with no lag or dropped frames over a VPN connection to my home file servers, if the internet connection where I happen to be is fast enough.

 

[ddrueding]

Youtube does offer 1080p streams, ddrueding.

I actually did point that out. My point being that most videos don't have 1080P versions yet, and most users don't even know about the quality drop-down yet.

 

[LunarMist]

I can play video files with no lag or dropped frames over a VPN connection to my home file servers, if the internet connection where I happen to be is fast enough.

Not in KY? :zzz:

 

[Howell]

Not in KY? :zzz:

You in KY? I'm in LEX!

 

[LunarMist]

No, but I thought Merc was. :cyclop:

 

[Mercutio]

I was in New Albany, immediately across the river from Louisville, for a month. It was crappy there.

And I could stream video from home when I was in the office where I was working, but not in my hotel room with the sucky wireless connection and no active ethernet ports.

 

[LunarMist]

Is that the reason for the dearth of posts lately?

 

[LunarMist]

My upload speed is much less than her download speed. Is it always that way for consumer IP service?

 

[BingBangBop]

As far as I can tell, yes.

Consumer's use cable, DSL, and satellite all of which typically have slower uploads than down. Further, the normal pattern of internet usage for a consumer is that there is far more data downloaded than uploaded. Consumers consume the data, businesses generate most of the data.

 

[ddrueding]

Yes. All consumer-grade internet service is asymmetric; significantly more download than up. The two main reasons are that normal end-user stuff fits that profile and that download is much cheaper; the ISP can cache most of it locally.

 

[LunarMist]

I don't notice any significant issues with surfing and only download apps occasionally. E-mailing is the slowest process.

 

[sechs]

I can triple my tested and stated BW by simply running three independent threads.

This wouldn't have anything to do with PowerBoost, would it?

 

[BingBangBop]

This wouldn't have anything to do with PowerBoost, would it?

I would say yes, except that power boost is supposed to only last a relatively short while and this was lasting throughout the entire download and upload which were aprox. 7GB each and taking many hours to complete.

 

[blakerwry]

There are many ways to limit bandwidth. Depending on the way it's done, this makes sense.

DSL, cable, and analog modems all listen, test, and grab parts of the frequency spectrum available on the wire when they boot up or initiate a connection. This is the available bandwidth they have to work with. This is usually a matter of installation, wiring quality, the provider's network design, and possibly the equipment (DOCSIS 2 vs 3, G.DMT vs ADSL2).

After that process, cable modems will download a config file which then tells the modem to apply QoS and policing to the user's data. A 'hacked' cable modem ignores these values. Similarly, DSL modems have a policing mechanism established at the DSLAM, limiting the speed the device will communicate at and what encoding is used.

Further up the chain, QoS and policing can be applied at aggregation routers (or dedicated traffic shapers) based on PPP connections, IP flows, etc. This is how bursting (aka power boost) works. Subscriber modems are allowed to sync at 20Mbps, for example, but a policer is set to limit the data rate to 10Mbps. Bursting may be allowed for the first 1Mbyte of an IP flow or based on the previous data usage.

In Sechs' example, it sounds like Comcast is limiting by IP flow. In many of the networks I work with, this is performed on the circuit via cable modem, sub IP, or PPP connection. It's a lot more work (investment) to perform this type of policing on individual IP flows, but it may be necessary for someone like comcast who wants to limit p2p traffic while not affecting netflix or their own premium services.