Continuous rightsizing

Key role(s)


Making the practice of rightsizing continuous


Regularly resize the infrastructure for better IT efficiency

After implementing the practice "1O1 - Rightsizing instances and storage options", the idea is to make it continuous. You therefore let the tool(s) of your choice run continuously to bring up optimization opportunities and set up a process for analyzing, validating and planning changes.

High-level process of continuous rightsizing
High-level process of continuous rightsizing

As seen in 1O1 practice, a high degree of sophistication in rightsizing can be achieved by choosing "burstable" types that further optimize the performance/cost ratio, but this requires a good knowledge of your workload's consumption profile. It is therefore recommended that you leave it running for 3 months before starting a rightsizing operation again.

Of course, this should not prevent you from detecting problems with badly crafted instances on the fly and being able to "fix" them. There is therefore a distinction to be made between:

  • Correction of badly crafted instances
  • Quarterly (or biannual) review of new opportunities for savings
See previous card

Anomalies alerting and detection

ID 2R3

See next card

Storage rightsizing and data lifecycle management

ID 2O2