Is 1.000 Terabytes a Month a Lot?

Two companies have made some interesting network usage statistics available recently. In the chart in this post, Verizon states that in December 2008, they transferred around 4.000 terabytes of data through their network. Ed Candy of Three in the UK reported 1.000 terabytes for the same month in his presentation during the Forum Oxford Conference. Incredible numbers, but how much is that on a day to day basis?

Let's do some maths. 1.000 divided by 30 days is 33.3 terabytes per day. Further divided by 24 hours of the day, by 60 minutes and 60 seconds, this amounts to around 3.85 GBit/s. During busy hour, traffic on the Internet is around twice the average as per the DE-CIX Traffic statistics shown here. That would bring peak traffic somewhere near 8 Gibt/s.

Let's further divide that number by how many cell sites an operator has. In the case of Vodafone Germany, it's around 13.000, so let's assume Three in the UK has around 8.000 to make the number low. A throughput of 8 Gibt/s divided by 8.000 cell sites is 1 MBit/s. As a cell has usually three sectors, that number has to be divided by 3. Hence, the average sector throughput during busy hour is around 0.3 MBit/s.

That's an average value of course. In conference presentations I have heard that cells in urban areas carry a lot more than the average. That's both good and bad news. The bad news is that capacity limits will appear here sooner. The good news is the number of base stations that have to be upgraded once the capacity of the current buildout is reached by adding a carrier or to add more base station sites is limited.

So the numbers quoted in the presentations seem quite realistic to me and also indicate that there is still quite soom room left to breathe capacity wise.

2 thoughts on “Is 1.000 Terabytes a Month a Lot?”

  1. Interesting analysis with simplicity. Converting Bytes to Bits was missed though, which will yield an 8 times higher result than what you concluded.

  2. Hi Edward,

    thanks very much for your comment, you pointed me to a big mistake here! If the number was really 8 times bigger then the network would be overloaded so something had to be seriously wrong with the numbers. I checked the presentation again and it was actually 1000 and NOT 10000 TB a month. With that the numbers are ok again.

    I corrected the title of the post and the text accordingly. Thanks again!


Comments are closed.