Dec. 12, 2018
Of all the means available to AWS users who are looking to reduce their cloud spending, there is one that definitely ranks amongst the most effective – Reserved Instances.
A Reserved Instance is a reservation of resources and capacity for a specific duration (1 or 3 years). Instances are available at various levels of computing power and AWS users purchase them at contract prices, plus hourly rates. The reservations of these instances are region-specific (us-east-1, eu-west-2, etc.) and not linked to a specific Instance to computing time. Users purchasing Reserved Instances can choose between Regular Reserved Instances or Convertible Reserved Instances (which provide more flexibility for an additional charge).
The basic idea is that when users purchase reservations, they commit to paying for all of the hours of the 1- or 3- year term; in exchange, the hourly rate is lowered. Details aside, what’s essential to know is that if used wisely, Reserved Instances have the potential to significantly lower your cloud bill.
Reserved Instances come in handy when AWS users know in advance that they’ll be needing a certain duration of computing power in the future. Purchasing Reserved Instances provides significant cost savings for the same amount of computing power, in comparison to new instances.
When purchasing Reserved Instances, users can opt for no up-front payment, partial payment or pay in full. The higher the upfront payment made by the user, the more savings (s)he accumulates. What follows is a table that perfectly illustrates the utility of Reserved Instances.
The different prices for 5 separate instances are outlined in the table below (RI stands for Reserved Instances):
|Instance||Quantity||Annual regular price||Annual RI price no upfront||Annual RI upfront 3 month||Annual RI upfront 12 month||Max Savings|
As illustrated by the table, by purchasing Reserved Instances, users have the potential to reduce their spending by up to 50%! Now that’s a significant reduction in spending, especially for companies who have relatively large footprints on AWS.
At the end of the day, it would be useful to note that it all boils down to one thing: accurately predicting the amount of computing power you’ll be needing in the future. This can be achieved by analyzing billing data for the past few months or simply by having the technical expertise required to make an accurate forecast of future needs.