Sky, like other internet providers, has an obligation to block piracy-enabling sites for a variety of reasons - be they court orders or injunctions handed down by the Premier League. Being a broadcaster, with a particular interest in getting people to pay for their sports, it's hard to imagine Sky would have any objection to this, but it's not always an easy task. So it's turning to analytics data to stay on top of things.
Blocking piracy sites has been likened to a game of Whack-a-Mole, or trying to kill a hydra. You kill one site and it'll immediately jump elsewhere, or worse it'll split off into multiple different iterations. So any tool that helps the company stay on top of the vast sea of internet piracy is going to be invaluable, and Sky claims using analytics and Google Cloud tools has already seen a "phenomenal reduction in pirate sites in the UK."
Sky UK CTO Mohamed Hammady spoke about this new strategy during Google Cloud Next in San Francisco earlier this week. He mentioned that Sky has been using NetFlow traffic information for “sampling the traffic on our core network,” a process that produced 500 billion data records within a single year. That's a lot of data, so the company turned to machine learning to help get through it all.
“Using BigQuery and an in-house algorithm – which cost $10,000 (£7,500) to develop – we are now able to continuously study traffic patterns with an always up to date list of suspect pirate sites. Once they have been confirmed as illegal they are shut down."
Apparently each query takes 30 seconds to achieve, and costs Sky $0.23/£0.18 a pop which Hammady claims is a good deal. It's also responsible for Sky expanding its relationship with Google Cloud for other data needs.
It's a bit odd to be hearing an ISP talking about their tactics in public, especially since pirates are always looking for ways to avoid being penalised and get around blocks set up by internet providers. But it's still interesting to know Sky is willing to be smart about its obligations, rather than just blocking random sites in a futile attempt to win the never-ending war. [ComputerWorld UK via TorrentFreak]