13.4. I/O Optimization

The existing Overview system used 40 o` files (named `o1-o40) to store overview data in an overview/ directory. To optimize I/O on a production server, spreading those files over multiple disks (or a RAID slice) was recommended. Tornado Front End uses the same concept, but automates the process by adding three additional tornado.conf directives:

TrackFilePath

What file system path Tornado Front End uses to locate the track file.

OverviewPath

What file system paths Tornado Front End uses to locate Overview files. Tornado Front End will use all the defined directories to optimize Overview file storage.

Headerpath

What file system paths Tornado Front End uses to locate Header files. Tornado Front End will use all the defined directories to optimize Header file storage.

OverviewPath /overview1,  /overview2,  /overview3,  /overview4,  /overview5
HeaderPath   /overview1,  /overview2,  /overview3,  /overview4,  /overview5

The track file is accessed frequently, but relatively small. Since the file will usually be in RAM, giving it a dedicated file system is unneeded.