Free software solutions for large distributed file systems


Storage has improved a million times in less than 25-years, more than Moore’s law for CPU speed. Companies need solutions to manage reliably the huge amount of data that is (or will be) stored in modern disk devices: free software is an answer to this problem. The talk will present the free software Gluster, which allows to build a distributed network file system that may scale up to the Petabyte scale on Linux (source: Facebook, internal communication at the Gluster Summit 2016).

We have been using Gluster for many years, and our largest installation holds 720TB of raw storage: our analysis show that Gluster based solutions are feature complete with commercial ones, and that leveraging standard hardware — instead of custom proprietary SANs — allows savings of more that 30% (or, dually, that you get more storage and performance for the same money).

Last but not least, these storage techologies are very mature and stable, so it is important to know that there are open alternatives to commercial SANs to build reliable storage systems.