AT&T rolls out 7.2 everywhere

[aristobrat]

No... Unfortunately, even though the towers can do 7.2Mb/s you still won't see the burst. Once the backhaul is up and running you will.

I guess I don't understand that part.

All towers currently have backhauls, it's just that the backhaul doesn't have enough bandwidth to make every data connection super fast when the backhaul is being shared by multiple data connections (iPhones, whatever). The more data connections to the tower, the slower each one goes, as the backhaul bandwidth is shared.

The current tower backhauls have to be faster than 3.6Mb/s. Otherwise, if 15 people were connected to that tower and using data, it'd be no faster than EDGE.

So if you connect to that tower at like 3am, and assume that nobody else is using data, I still don't see why you couldn't go faster than 3.6Mb/s. You wouldn't be sharing the backhaul (which is faster than 3.6Mb/s) with anyone else and the transmitters have been updated to now go faster than 3.6Mb/s, what would stop you?

 

[kdarling]

The current tower backhauls have to be faster than 3.6Mb/s. Otherwise, if 15 people were connected to that tower and using data, it'd be no faster than EDGE.

Many towers still have 2 T1s = 3 Mbps backhaul. A tower in a busy area might have 3 or 4 T1 lines = 6 Mbps.

A tower upgraded to 3G should have a couple more added, bringing the total bandwidth to ~ 9 Mbps.

That's not a lot these days. Planners now want ten times that much.

Carriers spend something like 25% of their operating budget on leasing backhaul lines... it's billions a year. Upgrading backhaul costs a lot.

Running fiber to a tower can cost $100,000.

That's why many carriers are going back and adding microwave based backhaul lines. (Half of European backhaul is microwave based.)

 

[jav6454]

Many towers still have 2 T1s = 3 Mbps backhaul. A tower in a busy area might have 3 or 4 T1 lines = 6 Mbps.

A tower upgraded to 3G should have a couple more added, bringing the total bandwidth to ~ 9 Mbps.

That's not a lot these days. Planners now want ten times that much.

Carriers spend something like 25% of their operating budget on leasing backhaul lines... it's billions a year. Upgrading backhaul costs a lot.

Running fiber to a tower can cost $100,000.

That's why many carriers are going back and adding microwave based backhaul lines. (Half of European backhaul is microwave based.)

Pretty much this... Except carriers don't use T-1 lines anymore, more of T-2s. I couldn't think of an easy way of saying it, hence, my little no so easy to understand post.

Think of it, as your home internet connection. Say you have a Wi-Fi router that connects your MacBook to the Internet. The router is 802.11n which is 600Mb/s (7.2Mb/s in a Cell tower). However, your speed will depend on how much you pay your ISP for. In this case say 10Mb/s. So, even though you have a 600Mb/s over the air connection, you won't see it since your ISP (or backhaul) is only 10Mb/s wide.

In a tower is pretty much the same deal.... except each tower has like 50 to 100 Mb/s which are divided among users of said tower. AT&T needs (minimum speeds) 1Gb/s speeds in the base towers for 4G and 200Mb/s for faster HSPA+ speeds they said they will implement given the average users.

 

[smithrh]

Pretty much this... Except carriers don't use T-1 lines anymore, more of T-2s.

Not sure where you got that idea. I've never actually seen a T2 provisioned in a cellular environment, ever. T2 does exist, I do know that, but actual usage in the field just doesn't happen.

T1s are nearly universal in North America, E1s elsewhere, with JT1 in Japan (not all that common anymore).

Even when backhaul is shipped over fiber or microwave, many times it's channelized in T1 format. Only recently has base station gear started to migrate to packetized backhaul with Ethernet connectivity for Metro Ethernet, which is where much of this is headed.

 

[ToroidalZeus]

Also I've noticed with AT&T is I can do a speedtest at one point in the day and get an amazing speed. Then, a few hours later, another speedtest gives me slower than edge speeds even though my signal says full 3G. Verizon might not peak as fast as AT&T, but the speedtests are consistent through out the day at least.

Thats because iPhone users are data whores. It's like cable internet. When there are a lot of users using the bandwidth at the same time it will be slower.

As for verizon, they dont have enough android smartphone users to experience this problem yet.

 

[kdarling]

Thats because iPhone users are data whores. It's like cable internet. When there are a lot of users using the bandwidth at the same time it will be slower.

It's not just the iPhone. (And web browsing isn't that intensive an activity.)

AT&T has recently said that most (over half) of their data congestion is from other feature phones with instant messaging apps.

They also said that switching to lower frequencies has allowed more people in buildings to get connections, and this had tripled their voice traffic in places like NYC.

As for verizon, they dont have enough android smartphone users to experience this problem yet.

Still, they've had millions of enterprise smartphone users for years, I believe a majority of RIM users, and quite a lot of laptop broadband users (which are the most data hungry type).

 

[Michael CM1]

I live in Atlanta just a mile south of the Georgia Dome and my speed test was underwelming. My download was 1.12mb and 6.9mb on my wifi network. I think that AT&lame need to find some faster mice for their exercise wheel. Apple may have the best smartphone but the provider is the worse in both performance and customer service.

I live up near Winder and just got 589 max with 222ms latency. WOOO. It's much better when I head into Lawrenceville. Thank goodness I'm getting about triple my EDGE speeds because I'd be pissed to have paid the $300 for a 3GS and not be getting good speeds.

Of course my 3Mb DSL connection makes that moot around here. But consider yourself lucky to not have REAL dead zones around you. I've got a couple in both directions that last for maybe half a mile. Hopefully no deer play chicken with me there.

 

[ToroidalZeus]

Still, they've had millions of enterprise smartphone users for years, I believe a majority of RIM users, and quite a lot of laptop broadband users (which are the most data hungry type).

Blackberrys and most other smart phones are not comparable to the iPhone. Most people who use BB are doing so for either the email, text msg, blackberry messenger or maybe yahoo messenger, anyways i think you are getting the point. They use them for text based services which use a TINY amount of bandwidth. iPhone users on other hand browser full HTML webpages with images. Stream youtube videos, stream radio stations. Download large game apps. Oh and do the email and text message stuff as well

As for USB or PCMCIA cards. No ordinary person pays for the added line just to use stream youtube on the go. Those are used primary by people who have a point i.e work related.

 

[kdarling]

I agree, YouTube streaming could be a cause of congestion from iPhone owners. Otherwise, I don't think iPhones do much realtime data.

Non-iPhone owners owners are far more likely to be background Pandora streamers on any carrier.

Non-iPhone owners are also the only ones who'll be major users of Sling or Skype over 3G on any carrier.

Broadband laptop users might not stream video, but they transfer large files.

The point is, ATT themselves have stated that the iPhone isn't their major data hog. And that's understandable, because they've made sure that Apple didn't allow the most heavy use over 3G.

It's not the load. A good network should handle that without dropping calls. It's the lack of ATT's network preparation for the load. That's what ATT themselves have said.

Cheers!

 

[ToroidalZeus]

Non-iPhone owners owners are far more likely to be background Pandora streamers on any carrier.

I didnt realize my razr2, vx8600, vx8300, sch-a870, etc. had the pandora app.

Non-iPhone owners are also the only ones who'll be major users of Sling or Skype over 3G on any carrier.

I didnt realize my razr2, vx8600, vx8300, sch-a870, etc. had the skype app.

Broadband laptop users might not stream video, but they transfer large files.

I didnt realize when you VPN into a network to configure a server you are transferring large files.
I also didn't realize when a business man is sending a spreadsheet of some sort it's a large file.

Plus very few people use usb/pcmcia wireless cards.

 

[kdarling]

I didnt realize my razr2, vx8600, vx8300, sch-a870, etc. had the pandora app.

Eh? We're talking about the heaviest data users. That usually means feature and smart phones.

In ATT's case, their CTO recently said:

While most people would assume that most of the wireless data traffic growth on AT&T's network comes from the iPhone, AT&T's executives said that isn't the case. De la Vega said that quick-messaging devices are actually driving a significant portion of data usage on the network.

"We have seen unprecedented growth on our network in the past couple of years," he said during an interview on the sidelines of the conference. "And the iPhone has certainly played a role. But it's not the only device driving growth. We have a lot of people going from basic feature phones to quick-messaging devices and other smartphones, which is driving data usage."

 

[thelatinist]

I played around with SpeedTest last night at home. On 3G I was getting decent downstream (~1300) and upstream (~230) but my ping times averaged in excess of 3500ms. Yes, 3.5 seconds to ping! I tried web browsing, and it was excruciatingly slow. Glad I have WiFi at home, but I can't imagine what it would be like for anyone relying on 3G data.

 

[kdarling]

Yes, 3.5 seconds to ping! I tried web browsing, and it was excruciatingly slow. Glad I have WiFi at home, but I can't imagine what it would be like for anyone relying on 3G data.

ATT's lousy ping times are a topic of great discussion on techie forums and blogs.

If you like tech reads, skim through this blog and the replies.

Some cell engineers make interesting comments, like this one in part:

AT&T are rather deaf to the best practices of the other 750 GSM/UMTS carriers around the world.

As a former GSM/EDGE engineer involved with the standardisation of the radio layer I am sometimes stunned that it works at all.

They don't apply the knowledge gained in other countries for buffer size and system parameters. The standards deliberately don't give guidence as to the default values (it's a European thing) but the vendors certainly do.

And even someone who works in that section for ATT chimes in, saying their routers are not being set up correctly.

We contract to carriers and see this a lot: the most capable engineers have retired or been laid off over the past ten years. So inexperienced folk are now doing much of the work.

 

[aristobrat]

I played around with SpeedTest last night at home. On 3G I was getting decent downstream (~1300) and upstream (~230) but my ping times averaged in excess of 3500ms. Yes, 3.5 seconds to ping! I tried web browsing, and it was excruciatingly slow. Glad I have WiFi at home, but I can't imagine what it would be like for anyone relying on 3G data.

Which SpeedTest are you using?

I tried the free Xtreme Speedtest iPhone app, and it shows the latency on my tests at only around 158ms.

 

[ToroidalZeus]

The point is, ATT themselves have stated that the iPhone isn't their major data hog. And that's understandable, because they've made sure that Apple didn't allow the most heavy use over 3G.

It's not the load. A good network should handle that without dropping calls. It's the lack of ATT's network preparation for the load. That's what ATT themselves have said.

Did you even read your own quote?

And the iPhone has certainly played a role. But it's not the only device driving growth.

All he is saying is that other devices are also growing.

 

[thelatinist]

Which SpeedTest are you using?

I tried the free Xtreme Speedtest iPhone app, and it shows the latency on my tests at only around 158ms.

I generally use Speed Test from Ookla/Speedtest.net. It's always given me good results. The latency issue has nothing to do with the software on my end; I consistently get ping times under 30 ms on my WiFi/Cable. At work just now I pinged about 150ms. The problem is on AT&T's end. It's not usually quite that high, but it's consistently over a second even on 3G.

 

[strike1555]

As for USB or PCMCIA cards. No ordinary person pays for the added line just to use stream youtube on the go. Those are used primary by people who have a point i.e work related.

So do you have any sources or do you just like to pull facts out of your ass?

 

[ToroidalZeus]

So do you have any sources or do you just like to pull facts out of your ass?

What sources do you need? Use your eyes.

 

[alent1234]

I guess I don't understand that part.

All towers currently have backhauls, it's just that the backhaul doesn't have enough bandwidth to make every data connection super fast when the backhaul is being shared by multiple data connections (iPhones, whatever). The more data connections to the tower, the slower each one goes, as the backhaul bandwidth is shared.

The current tower backhauls have to be faster than 3.6Mb/s. Otherwise, if 15 people were connected to that tower and using data, it'd be no faster than EDGE.

So if you connect to that tower at like 3am, and assume that nobody else is using data, I still don't see why you couldn't go faster than 3.6Mb/s. You wouldn't be sharing the backhaul (which is faster than 3.6Mb/s) with anyone else and the transmitters have been updated to now go faster than 3.6Mb/s, what would stop you?

those numbers are ideal lab condition numbers. and even in the real world you would have to be in an ideal position in relation to the tower to get that performance.

from where i sit at work, i can get a difference of a few hundred kilobits in a speedtest just by moving my phone around 12-36 inches

 

[alent1234]

ATT's lousy ping times are a topic of great discussion on techie forums and blogs.

If you like tech reads, skim through this blog and the replies.

Some cell engineers make interesting comments, like this one in part:



And even someone who works in that section for ATT chimes in, saying their routers are not being set up correctly.

We contract to carriers and see this a lot: the most capable engineers have retired or been laid off over the past ten years. So inexperienced folk are now doing much of the work.

i'm shocked

i'm in NYC and when i run speedtest it shows my external IP as being just north of Wichita

 

[aristobrat]

I generally use Speed Test from Ookla/Speedtest.net. It's always given me good results. The latency issue has nothing to do with the software on my end; I consistently get ping times under 30 ms on my WiFi/Cable. At work just now I pinged about 150ms. The problem is on AT&T's end. It's not usually quite that high, but it's consistently over a second even on 3G.

Oh, I know it wasn't the software causing the latency, I just wanted to be able to test using the same software (for comparison). Ping times in my area are a lot lower than yours.

 

[aristobrat]

So far over the last week my fastest down speed was 3789kbps @ 3:25pm on Saturday the 2nd. Obviously a lot less business traffic on the towers and system during a Saturday.

those numbers are ideal lab condition numbers. and even in the real world you would have to be in an ideal position in relation to the tower to get that performance.

I was just curious about exceeding 3.6Mb/s before the backhual was upgraded. From the quote above, it does look look like it's possible.

 

[thelatinist]

Oh, I know it wasn't the software causing the latency, I just wanted to be able to test using the same software (for comparison). Ping times in my area are a lot lower than yours.

I would be very happy to get that kind of ping time.

 

[kdarling]

We do some enterprise Blackberry apps in NJ and NYC, and once we had a weird timeout problem. Had to get RIM's help digging into it.

Turned out that a NYC customer's packets were being sent from NYC to Canada all the way to Japan and back, due to some misconfiguration on their side.

Love this job. Never dull.

 

[Mischka07]

What is this 7.2 Mb/s that you speak of? I clock in at 252 Kb/s in San Francisco .