PostgreSQL vs. data corruption a talk by Tomáš Vondra

Data corruption is a major issue in databases, particularly in large and old ones, and tackling it may be surprisingly difficult. In this talk I’ll discuss where data corruption comes from – why and how does it happen, and what can we do to deal with it. Either by implementing new features in the database system, configuring it properly, using the right hardware, and various other means.

I’m a PostgreSQL developer and contributor, so I’ll focus on data corruption issues in this database system. But while I plan to discuss a number of PostgreSQL-specific examples, data corruption is a generic issue that applies to most regular databases. So hopefully the conclusions of the talk can be helpful for users of other databases.