Analyzing cacheable Traffic for FTTH Users using Hadoop

We present this year (2015) statistics about cacheable traffic in the access network of Orange in Paris for about 30,000 customers served by a fiber to the home subscription. These statistics update some of the results presented in a recent work, which considered only 2\,000 fiber users in 2014. The huge amount of data to be processed in the new vantage point made necessary the usage of a hadoop cluster that we have used to process the data and report new statistics in the present paper. The aggregation level at which we observe web traffic allows to draw some conclusions about the feasibility of implementing in-network caching at wire speed.