Database - Data Life Expectancy

The data life expectancy is the amount of time that data is expected to remain in the database. You need to establish the requirement for how long fact data needs to be held online at the fact level. Older fact data can often be held online in aggregated form, reducing the amount of space it requires. This is a useful trick, but it has a cost. Rolling up the data requires processing power, and can require more space in the short term. This needs to be designed into the warehouse manager’s processing schedule.

For archiving there are three figures that need to be established. First, you need to establish the start date; that is, when archiving is required to begin. For example, if the aim to keep 3 years worth of data online, and the data warehouse is to start off with 18 months of historic data loaded, you will need to begin archiving 18 months after the go-live date. If designed properly, the load process will not depend on the archiving process being run first. Sufficient space should be made available to allow the archiving to occur later than immediately required. This allows the archiving to be performed at a more convenient time.