Every now and then I run into a conversation with my clients around planning capacity for anticipated database growth. Although there are straight forward methods for calculating data/log file and object growth rates from catalogued views in the database which lead to linear predictions, there are several considerations that must be taken to make this exercise more appropriate in the real-world.
Figure 1 – Instance Growth Trend using Foglight for Databases
How much of the foreseeable future are you planning capacity for? Will your organization be archiving data or keeping all data? Depending on the business, there may be policies in place that require a certain amount of data to be retained. Other factors that affect capacity planning for database instances include accounting for additional users or growth in business volume. That is where knowing the current data growth as well as look at trends over a historical period of time, such as 12 months or the past year. The time might be constant, but the growth of data can grow exponentially.
Figure 2 – Database Capacity Planning Report using Foglight for Databases
Will the applications supported by the databases have newly added capabilities that affect the size and shape of your data? For example, the application may incorporate the ability to store image data or other sources of information which could dramatically affect database sizes. Will developers or DBAs be adding indexes through performance tuning efforts? These efforts can also have an effect which can amplify the size of the data by a factor of 2x or even 4x.
Figure 3 – Tracking new index creation over time using Foglight Performance Investigator
The key is having the right tools and asking the right questions which makes for a more informed decision around planning capacity around anticipated database growth. While we make these considerations, the goal is to also marry capacity planning with database performance monitoring. This combination we be mutually beneficial to both IT and the business it supports.