dslreports logo
    All Forums Hot Topics Gallery


how-to block ads

Search Topic:
share rss forum feed


reply to NormanS

Re: AT&T usage meter -- is yours as wildly inaccurate as mine?

Mine is completely broken this week - it appears that no data has accumulated for two weeks or something is otherwise wrong with the meter.

Calling AT&T and speaking to several people left me with a few very weird conversations about the future of the metering, and I'm really hesitant to post the things I heard AT&T reps say because some of the stuff was pretty out there. People are NOT happy with the caps or the metering.

I'm just going to send a polite certified letter to the OOP and see what (if anything) happens to fix my meter.


reply to gvd

Funny, caps have been in place since May 2, 2011 and Texas can't see their usage down here. "Under construction and don't worry about it" the page says.

It's coming
Mineral Ridge, OH

"Maintenance In Progress

AT&T is continually working to improve our internet site for our customers. As part of this ongoing effort, we are currently doing maintenance. We apologize for this inconvenience and ask that you please try back again later."

What a joke.
Yeah, it's kind of like that.


said by dslwanter:

"Maintenance In Progress

AT&T is continually working to improve our internet site for our customers. As part of this ongoing effort, we are currently doing maintenance. We apologize for this inconvenience and ask that you please try back again later."

What a joke.

the cap is in place and u're scared
scared of more pay?
scared of meter bar?
you're so funny
baaa baaa baaa!


Cleveland, OH
reply to gvd

I find it disturbing that the usage meter for my account shows almost 10 GB of data for the two week period of May 12th - May 26th since during this period the connection speed was often 224k due to the bad wires. It's nowhere near the cap but even so I think it shows the meter is inaccurate. There's no way I was transferring almost 1 GB per day with a 224k connection. A large portion is probably retransmission of data lost due to dropped connections and noise. With this metered data use increased line noise means more lost data and more data transferred.



We just blew past our 150 limit this month, the past two months though we've been significantly under. This month though since all the TV shows have been ending, we've been watching a lot more hulu and netflix streaming.

I installed Netlimiter on my HTPC to see how much bandwidth it used to stream a movie and we watched one. It came in at 1.3GB for 90 minutes. So that's going to be a significant impact.

However I also have no doubt that ATT is over billing. Or rather from a consumers point of view we're using X amount and from their point of view we're using Y amount because they're taking into account re-transmissions because of their crappy network and ATM/PPoE overhead as others have already said. There could also be inaccuracy in their metering as well.

There already is a class action lawsuit goign on against ATT for their caps. There's no reason why DSL is locked at 150 when the same 6mbps service on U-Verse internet only service gets 250. They're using it to force heavy bandwidth users into a different service. So hopefully this lawsuit goes someplace. If they are overbilling though, there's no doubt the techies out there are going to uncover it. We don't live in a day and age where they can get away with that kind of stuff anymore, but i'm sure they'll try to until they get their knuckles slapped by the FCC or someone.



When cable started placing caps some 5 years ago, they were at 250 GB. For AT&T to cap at 100 GB lower they must be saying that their infrastructure is much worse than cable was back then.

My modem and router (separate pieces of hardware) have data meters that are showing very similar amounts. Yet AT&T's is consistently higher by some 15-20%. They need to be open about what data they are collecting, how they are measuring it, and where in the infrastructure they are measuring. Openness is the only way their customers will trust them.


reply to gvd

My usage meter has been stuck since the 26th. What bugs me, besides the cap is that it's four days behind, so you can't really judge how much Internet you're using. For instance if I'm at 132g at 5/26 then it's 5/31 I can't really tell how much I've downloaded.

I can keep track myself, but my usage meter is way off compared to what I figure. I can't figure out how I uploaded half a gb when my computer wasn't even on for one day. I guess it must be timing. Perhaps they don't use a midnight to midnight range or something



Right. Certainly, countless users would be satisfied if AT&T was open about what they are doing. Give an honest explanation about the variances that we are seeing. Explain what, where, and how. Don't hide it. Will they be open? I'm not holding my breath.


Elmhurst, IL

Open?! AT&T... NEVER. They lie, cheat and steal with every breath, Ma Bell is Back with a Vengeance !
I got the email about the caps, tried it once - seemed high. Today I tried again, same address they supplied and... NOTHING! Get an error message:

Current High Speed Internet Data Usage Details

USG310: We are sorry, but we are unable to display your Internet usage.You must access this tool using your primary AT U-verse or DSL account.Please try again.

Is that BS or what. Forget support - they heard you to a useless forum that would be more at home on Facebook. Your other option is live chat (right) or CALL THEM - and wait 45 minutes for someone to answer.

It's getting worse folks, and our options for HS DSL are getting smaller. Now, just where can I go?

I gave her time to steal my mind away
San Jose, CA
·Pacific Bell - SBC
reply to tekwyzrd

said by tekwyzrd:

There's no way I was transferring almost 1 GB per day with a 224k connection.

224 kb/s? Have you actually done the math?

60 seconds to the minute.
60 minutes to the hour.
24 hours to the day.

Your multiplier should be: 60*60*24 ... comes to 86,400 seconds.

19,353,600,000 bits per day? Is that more, or less than 1 GB.
~Oh Lord, why have you come
~To Konnyu, with the Lion and the Drum

reply to gvd

For all those complaints about this at least they are counting 1GB = 1024MB and 1MB at 1024KB vs. 1GB=1,000,000,000B ; this is per the historical usage page. If you do the math, for the 150GB/month plan that amounts to 11,061,273,600 bytes additional you get by calculating this way. And for the 250GB/month plan it amounts to 18,435,456,000 additional bytes per month.


Batavia, IL
·AT&T Yahoo

It's ironic that what you imply as generosity on AT&T's part is less than the amount that AT&T is over-reporting on my account each month...

This cycle - 5GB difference so far between AT&T and dd-wrt
Last cycle - 14GB difference between AT&T and dd-wrt
Prior cycle - 16GB difference between AT&T and dd-wrt

And that's without going over the cap for those billing cycles.


Well just because their website says it, doesn't mean they do it - for that to happen it would mean that their marketing types would have to talk to their software guys who wrote the program. But your post is coincidentally interesting nonetheless.

dave sz

Carpentersville, IL
reply to gvd

Can someone explain to me why the status page is different every week? My login info no longer works as when I enter the user name I used before it now says I need to enter the cell number. Of course when I do that it says the cell is not authorized to see this info.


said by dave sz:

Can someone explain to me why the status page is different every week?

Because it's AT&T and they can and they do.


Buffalo Grove, IL
reply to gvd

My meter has been dead for about a month now. Just stopped updating, and when the new and "improved" one came up, still no joy. I assume the usage collection is continuing, because after the the summary stopped updating, I could still see the daily data.

Now, with the cool new page, I can't see any detail usage, and I get AT&T ads...yet for some reason I feel no desire to buy more of this quality AT&T service... (`_`)

dave sz

Carpentersville, IL
reply to gvd

Finally works again. Still 4 days behind. Still useless for the last week of every cycle as I'm usually pretty close to the limits.


said by dave sz:

Finally works again. Still 4 days behind. Still useless for the last week of every cycle as I'm usually pretty close to the limits.

I believe the 4 days behind is for security purpose and by design. The electric company does the same with online power meters.

dave sz

Carpentersville, IL

Security purpose? How so?
It just makes those close to the limit paranoid.
Either way, checked it now to see what my usage was in the last 4 days of the cycle, and it looks like none of it was counted or the data they showed up to that point was incorrect. I was at 109gb with 5 days left and it shows 109gb after the term ended.


Chicago, IL
reply to bbear2

Only one PC here and no way to see bytes from router or modem so I wrote an app (in VB10) that gets the bytes sent/received per second from my LAN card.

In order for my app to match what ATT is saying I'm using the app divides the daily usage byte values by 980 instead of say 1024 or 1000. I guess its due to overhead/re-transmissions, or they can't add properly?


Wheaton, IL
reply to gvd

Click for full size
I think mine is broken. As you can see it went down a lot, although the 2 60gb months are what I would consider average usage for me. I don't see how 20gb can be accurate for July, and my usage for August is currently at 0.00 gb, which also can't be accurate because I started streaming more video once I saw that I had some overhead on the cap. Anyone else's like this? Seems like it happened around the time they changed the usage page.


reply to gvd

I have also been trying to figure-out what exactly is going on with the usage meter. And I think I have an answer for at least some people. The short answer, if you have older dsl(not Uverse) in most cases this dsl is using ATM(Asynchronous Transfer Mode) to transport the Layer 2 PPPoE between the DSLAM and your DSL modem/router. This adds about 20% overhead to your Internet Traffic. AT&T seems to be measuring the raw usage including this overhead which is why for some people the numbers on the usage meter are off by about 20%.

I would like to point-out that this extra overhead only exists between the DSLAM and the DSL modem/router. To give you an idea just how much this effects things... those of us with older DSL where AT&T is including this ATM Overhead in the usage meter means that we hit our 150GB data cap after only about 120GB's of actual Internet traffic which hardly seems fair to me.

The Long answer with detailed data and examples follows. I am going to assume that you if are still reading this you want to know why exactly this is happening. Before I dig into the gritty details, some background information is needed. To transmit data on the Internet all data is broken into IP(Internet protocol) packets. When using Ethernet the maximum amount of data in an Ethernet frame is 1500 bytes. When using PPPoE( PPP over Ethernet) you have to add a 8 byte header to the IP packet. So this limits the maximum IP packet size to 1492 bytes. The Ethernet frame also has 18 bytes of header. The math will get complex below but if you don't know these basics don't think you will follow the stuff below...

Since traditional DSL uses ATM as the transport; ATM breaks data into 53 byte cells(basically very small packets). Each cell has a 5 byte header and 48 bytes of data. The total overhead of using ATM is complicated by a more than just this header on each cell. Each PPPoE packet has to be encapsulated in what is called a LLC/SNAP header(Also known as RFC1483 encapsulation). This SNAP header adds a 8 byte header to the packet. But it doesn't end there because SNAP requires AAL5. AAL5 adds a 8 byte trailer to the last 8 bytes of the last cell in the AAL5 packet. After all of that you also have to add the 8 bytes of overhead for running PPPoE. And the 18 bytes for the Ethernet header. This total packet is called a PDU in ATM networking terms. It's this PDU that gets split up into ATM Cells. And each cell can carry only 48 bytes of the PDU.

So this is quite a bit of overhead per IP packet... lets add it up.

5 bytes of ATM header for every 48 bytes of IP data + 8 bytes of SNAP/llc header + a 8 byte aal5 trailer in the last ATM cell + any needed padding in the last ATM cell + 8 bytes for PPPoE + 18 bytes of Ethernet header = total minimum overhead is 29 bytes.

In reality, even this isn't correct because the minimum IPv4 packet size is 20 bytes so when you add up all of the overhead you would need to have at least 2 ATM Cells(one of which would have some padding in it) to carry packet to the DSLAM.

So since packet size effects the amount of overhead ATM causes let me give you a few examples... a 64 byte IP packet, a 578 byte packet and a normally max size(if you are running PPPoE) 1492 byte IP packet.

--- 64 byte IP packet ---

64 byte IP packet + 8 byte PPPoE header + 18 byte Ethernet Header + 8 byte SNAP/LLC header + 8 byte AAL5 trailer = 106 bytes to be transmitted via ATM Cells.

106 bytes / 48 bytes = 2 with a remainder of 10
This means we need three ATM cells to carry this data. The first 2 cells will carry 48 bytes of data each. But the last cell will have 10 bytes of data(2 bytes + the 8 byte AAL5 trailer) and 38 bytes of zero padding.

So for these three ATM cells we will have 15 bytes of ATM header + 38 bytes of padding.

Total data usage to transmit a 64 byte IP packet is 159 bytes on the ATM Network. Or 148.4% overhead.

-- 578 byte IP Packet --

578 byte IP Packet + 8 byte PPPoE header + 18 byte Ethernet Header + 8 byte SNAP/LLC header + 8 byte AAL5 trailer = 620 bytes to be transmitting via ATM cells

620 bytes of PDU fit into 13 ATM cells(65 bytes) + 4 bytes of zero padding.

So a 578 byte IP packet transmitted to the DSLAM uses 689 bytes total. Or 19.2% overhead.

-- 1492 Byte IP Packet --

1492 byte IP Packet + 8 byte PPPoE header + 18 byte Ethernet Header + 8 byte SNAP/LLC header + 8 byte AAL5 trailer = 1534 bytes to be transmitted via ATM cells.

1534 bytes of PDU fit into 32 ATM cells(160 bytes) + 10 bytes of zero padding.

So a 1492 byte IP packet transmitted to the DSLAM uses 1702 bytes total. Or 14.0% overhead.

So you can see that as the packets get larger in general you have a lower overhead percentage. In the real world most you tend to have a mix of large data packets and small reply packets. So for most Internet traffic you see about 20% overhead from using an ATM transport.

In my case I have been monitoring my usage since September 27'th which is when the month rolls over for me. I have been using SNMP to monitor traffic usage on the Ethernet interface connected to my Speedstream 5100 Dsl modem(running in bridge mode).

I went on Vacation on October the first which gave me a nice a clear line of no activity. This allows me to look at great detail the first 4 days of usage. I intentionally used greater than normal usage on the 27'th and 30'th just so I could show bigger numbers. I expected to see about a 20% difference but I wanted to difference to be Gigabytes not just a few hundred Megs... So here are my raw numbers + formatted numbers.

Raw data counters on 9-27
IF-MIB::ifHCInOctets.6 = Counter64: 165559800409
IF-MIB::ifHCOutOctets.6 = Counter64: 24971968830

Raw data counters on 10-7
IF-MIB::ifHCInOctets.6 = Counter64: 176707635191
IF-MIB::ifHCOutOctets.6 = Counter64: 26849034117

total for the month so far.

Download 11,147,834,782 bytes = 10.382 GB
Upload 1,877,065,287 bytes = 1.748 GB
Total 12.13 GB

AT&T shows...
Down Up Total
10/07/2011 0.01 0.01 0.02
10/06/2011 0.01 0.01 0.02
10/05/2011 0.01 0.01 0.02
10/04/2011 0.01 0.01 0.02
10/03/2011 0.01 0.01 0.02
10/02/2011 0.01 0.01 0.02
10/01/2011 0.01 0.01 0.02
09/30/2011 4.20 0.53 4.73
09/29/2011 0.54 0.19 0.73
09/28/2011 0.79 0.21 1.00
09/27/2011 6.47 1.38 7.85
Totals 12.07 2.38 14.45

So they say I used 14.45 GB this month when I really only have used 12.13 GB total. In other words, it looks like they are adding about 19.12% overhead to every Gigabyte I transfer.

This overhead is only on the ATM connection between my DSL modem and the DSLAM. I don't feel that it's legitimate to count this traffic that doesn't go any father than that. For instance... all of the above stuff about ATM overhead isn't valid for Uverse service as it uses VDSL and they don't use ATM as a transport between the DSLAM and VDSL modem/router. This means that the raw numbers from the DSLAM would only be off by 1-2% from the Internet IP traffic from a customer. Nowhere on AT&T's website does it say that if you have older service expect to receive a 20% surcharge on the usage meter. Nor does it say that you really can only transfer 120GB of traffic from the Internet and not 150GB as they claim.

For other people where the usage numbers are off more or less than ~20% I can't explain that. It is possible that AT&T has data collection problems for some users. But at least in my case the numbers do look correct if you assume they are also counting ATM overhead.



dude I said this 7 months ago!

»Re: AT&T usage meter -- is yours as wildly inaccurate as mine?

They are totally counting ATM transport....it's major BS!

Reno, NV
reply to gvd

Something funny showed up when I made my monthly check to see if AT&T was measuring my usage yet:

Current High Speed Internet Data Usage Details

Note: You do not have DSL Internet service on this account, so there is no usage to measure.
They certainly know that they must bill me for my DSL service and that monthly charge is shown elsewhere on their management site. So I guess I have unlimited DSL bandwidth by default.
TCE Weather