Designing for Backups
There was a time not so very long ago when backups were a minor issue in database management. You set up a batch job to perform the backup overnight, while the system was not being used by anybody else, and that was that. Several issues have complicated that once common scenario.
First is the tremendous increase in the size of databases that can be supported by a relational database management system like Teradata. Where at one time a “very large database” might be on the order of several GB, it is increasingly common to see multiterabyte databases supporting large data warehouses and support for petabyte and even yottabyte databases is on the near horizon. Magnetic tape is a serial medium. Even if you archive to multiple tape drives simultaneously, there is still a considerable time issue involved with backing up multiterabyte databases.
Second is the end of the era of the batch window. Enterprise data warehouses now typically support worldwide operations, and that means that the system must be available to users 24 hours per day, seven days a week. Without the old batch window, how do you back up your business-critical data without having a significant negative effect on the capability of your data warehouse to support its worldwide user base?
Many approaches to solving these problems fall outside the scope of database design, but at least one method for minimizing the impact of backing up very large databases can be designed into your database. That method is the subject of this topic.