Database systems are under pressure to become more and more powerful. But reliability seems to be suffering as a result. Now, ETH computer scientists have developed a tool to automatically detect logic errors in database systems using three different methods. They found and reported over 450 unique, previously unknown bugs.
When a smartphone stores an address, we assume that we can find the address again using the right query. Just like all matching items in an online store can be listed using the right search terms. There’s just one problem: it’s not so much a matter of course as we believe. A logic error in the database management system (DBMS), which manages the stored information, can return incorrect responses without us even noticing the error.
A surprising number of errors in all systems
These types of logic error are surprisingly common. Or rather, they were until recently. Researchers at the Advanced Software Technologies Lab of ETH Zurich have now developed a tool that automatically detects logic errors in database software. It has already found over 450 programming defects in widely used database management systems.
Source: “450 logic errors found in popular databases”, Daniel Meierhans, Zurich ETH News