Compute Optimisation - methods of reducing cloud costs

Scheduling - the process of defining when compute needs to be running

This cost optimisation method is actually the method of achieving the highest savings and cost reductions. Having applications running ( thus consuming compute infrastructure ) only during the business week working hours of 8am to 6pm for example, actually means that those same compute needs are not running for 70% of the full week - against the 24 by 7 equivalence. Thus compute running just 50hrs of the 168hr week = just 29.7% compared to running 24*7.

Cloud costs are based upon compute runtime ( for server instance and database instance ) - where stopping and therefore not running outside of the standard Monday to Friday working hours, means that no charges are generated - hence the 70% cost savings.

This can be achieved by having application platform teams manually start and ( at end of day or when finished using ) stop their systems ( best suited for development environments ). Or by scripted automation to achieve begin of business day start-up, and another script for end of business day stop.

Cloud monthly opex spend savings of 70% - of your compute and database instance bills. Put into perspective, your monthly cloud bill is roughly of four roughly equal parts - compute spend, database spend, storage spend, and everything else added together. So application of the cost savings method of scheduling is targeting 50% of your monthly bill!

As I said ideally suited to cloud consumption needs and thus reducing cloud opex costs. But applying the same “run it only when you need it” against legacy data centre infrastructure, will ensure virtual machine demand will be kept efficient, and where large development and test work takes place, only running dev and test environments when needed, and better still, tear down when no longer needed, ensures efficient infrastructure use, which should keep total demands lower leading to less capital spend over time.