Not long after AT&T confirmed that it would begin to implement bandwidth caps on DSL and U-Verse customers beginning in May, some of the more technically savvy customers began to notice that the bandwidth meters supplied by AT&T for customers to check usage online are reporting major discrepancies in data usage compared to customer router logs. A detailed explanation of AT&Tâ€™s current measurement methodology follows below:
â€œThis is the problem. DSLAM’s are layer 2 devices in the OSI model. Unlike Uverse which uses IPDSLAM, the counters for an ATM based DSLAM are going to include the PPPoE overhead and ATM overhead. Conservative estimates would put PPPoE and ATM overhead at 30%. Of course it all depends on the type of data sent and the size of the packets. If AT&T is indeed pulling it’s usage statistics from the DSLAM then they are billing customers for the overhead injected by their use of PPPoE (1500 – 8 = 1492) and for the overhead necessary to transport a packet over ATM (due to the minimum 53 byte atm cell size).â€
This means that AT&Tâ€™s own tools for data usage measurement are counting the overhead bits necessary for transferring data from the switch to the router, which would explain the sudden increase in data usage from one month to the next. Other reports also make note that depending on the date, the same tools will under report data usage by as much as 80% of actual usage compared to router logs, which could lull unsuspecting customers into a false sense of security about actual data usage and setting up a situation where overages can occur, making the next bill much more expensive than expected. AT&T has so far not commented on the reports.