Testing AT&T u-verse

[MichaelB]

AT&T u-verse was just installed. The idea is to replace our comcast broadband, which has great speed but poor reliability. We are at the outer limit of u-verse availability and the signals needed to be boosted, so I would like to test the capacity, and my plan is to simultaneously engage skype, voip, neflix HD progam, and then upload a dozen or so 1MB files. The upload speed offered is 1.5mb, so it should be good, and I'll run a speedtest while all the is happening. Any other suggestions?

 

[REWahoo]

Maybe stand on one leg and face north while you do it?

Funny, while you posted this I was on the phone talking to my ISP about upgrading to faster $ervice...

 

[Rustward]

speedtest.net does a pretty good job of determining your bandwidth capacity.
It sounds like what you really want to do is determine how all that traffic will play together at the same time.

pingtest.net does an excellent job of determining your connection quality. You can have a boatload of bandwidth and also have crappy quality.

I would run both speedtest.net and pingtest.net (serially) while those other loads are running.

 

[modhatter]

Can you explain what the Ping test numbers indicate. I understand that the higher numbers are not good. Is that correct. What is considered good? As an example. My download speed is 9+ and upload is 3.7 and ping is 17 ms.

Would love to hear how the U-verse works out for you after a few weeks.

 

[Bestwifeever]

I'll be interested in your results, Michael. U-verse is bombarding us with offers, but I don't think they're any less expensive than Comcast overall.

 

[MichaelB]

Can you explain what the Ping test numbers indicate. I understand that the higher numbers are not good. Is that correct. What is considered good? As an example. My download speed is 9+ and upload is 3.7 and ping is 17 ms.

Would love to hear how the U-verse works out for you after a few weeks.

Ping is a test that determines latency - how long it takes to send a signal from your computer to a network server. Your numbers look pretty fast to me...

I'll be interested in your results, Michael. U-verse is bombarding us with offers, but I don't think they're any less expensive than Comcast overall.

Just did it and the skype quality does not appear any worse. I'll try again in a day or two, but first I want to see it there are any service interruptions.

 

[ERD50]

Just did it and the skype quality does not appear any worse. I'll try again in a day or two, but first I want to see it there are any service interruptions.

Good quality VOIP is dependent on more than just speed (jitter is the big one). There are some VOIP quality tests out there. But I'll suggest you do your multi-loading as you described, and do a 'real life' test of the phone call:

Call a willing participant, and take turns counting steadily and robotically from 1 to 20. This way, if there is a delay or dropped audio, it is obvious. In normal conversation, it's tough to always know if the drop was the speaker hesitating or not.

-ERD50

 

[MichaelB]

Speedtests show skype consumes lots of bandwidth, but even with outgoing phone with vonage and lots of downstreaming (pandora & netflix) there's a bit of upload capacity left. Not great, but it will do if it proves to be reliable.

Feever, if you have Comcast and are generally satisfied there is probably no advantage to change. My brother just switched to U-Verse, but that was because of poor offerings by the only other alternative, a local provider in the west Chicago suburbs. We are changing because Comcast internet service is unreliable (I think our hub is overloaded) and they are too busy to come out to the house and even investigate.

Good quality VOIP is dependent on more than just speed (jitter is the big one). There are some VOIP quality tests out there. But I'll suggest you do your multi-loading as you described, and do a 'real life' test of the phone call:

Call a willing participant, and take turns counting steadily and robotically from 1 to 20. This way, if there is a delay or dropped audio, it is obvious. In normal conversation, it's tough to always know if the drop was the speaker hesitating or not.

-ERD50

I still need to improve the phone test, as you suggest. All I did was dial out and keep the line open, but did not test the voice quality. The next test needs to be more rigorous.

My speedtest results with the system fully loaded

 

[Nords]

... so I would like to test the capacity, and my plan is to simultaneously engage skype, voip, neflix HD progam, and then upload a dozen or so 1MB files. The upload speed offered is 1.5mb, so it should be good, and I'll run a speedtest while all the is happening. Any other suggestions?

Is your router messing with your load-test plan by prioritizing your bandwidth to favor one data stream at the expense of the others? I don't know if that router feature has to be selected or if it's a default.

 

[ERD50]

My speedtest results with the system fully loaded

I'm guessing that the 300ms ping could be a problem with VOIP. But that was fully loaded, so it might be fine under light loading.

The 'jitter' problem I referred to - in a VOIP (or cell phone), they take a snippet of sound, digitize/compress it into a 'packet' and then transmit that packet to the receiving end. Over the internet, each packet takes whatever path is available, it's all very dynamic routing. So a packet sent at one point will sometimes arrive later than one sent earlier. This isn't a noticeable problem on a web page. The page loads in a second or two, and you really don't care much what loaded first.

But with voice, if sequential packets of sound don't arrive in some reasonable time, they have to just drop that snippet. You can't wait a second for it to show up. This is one reason why there is a noticeable delay in VOIP (and cell phone if you listen for it). The delay is required for the time it takes to sample enough sound to compress, then transmit it and allow for some 'jitter' in arrival times of the packets, and then re-construct it all at the receiver.

It's a miracle this works at all!

Anyhow, I suspect that jitter is generally pretty closely related to long ping times. Or at least some X amount of jitter is going to be more of a problem if the ping time average is already high - the system can only wait so long.

I've got rather marginally 'high-speed' internet. At best I get 3Mbps down and 1Mbps up, and 20-30 mSec ping times, but it varies all over. My VOIP is fine 85% of the time, and the other 15% is up for grabs, from mild chopping to unusable and dropped calls. I use a local point-to-point wireless provider, which means I don't need to deal with a monopoly, which does more to keep my blood pressure in line than the meds I take.

-ERD50

 

[MichaelB]

Is your router messing with your load-test plan by prioritizing your bandwidth to favor one data stream at the expense of the others? I don't know if that router feature has to be selected or if it's a default.

I think yes, because I started up skype first and then added loads one at a time. The last load, uploading around 8m in emails, took forever while Skype stayed more or less constant. I have more testing to do, and also need to see what router is doing and what I can change.