The common reason for using simple file storage for a cache rather than a denormalized RDBMS table is that the file storage is much faster and consumes fewer resources. There's no network transfer, no SQL parsing, and no logging. And with the Linux file system hot files are generally kept in RAM, making the file system a sort of shared memory cache. It may seem less elegant, but the speed can't be beat and CPAN modules make it really easy.