We have been advocating the value of AWS Reserved Instances for some time and why not! They allow you to save a lot on the on-demand resource by committing to a low cost reservation early on. But, as the size of data being produced is increasing every day, real time data analysis and unplanned minor spikes in the resource requirement in a very small fraction of time need not hit the trigger to a Reserved Instance rather kickstart a small VM out of the whole resource.

Let’s look into the blog by Jeff Bar, the chief evangelist of Amazon Web Services, how he likes to explain it. If you owned an an RI for a c4.8xlarge, with this RI you may use any of a Linux/UNIX C4 instance:

  • One c4.8xlarge instance.
  • Two c4.4xlarge instances.
  • Four c4.2xlarge instances.
  • Sixteen c4.large instances.

Now, you could buy an RI for a c4.4xlarge, use that instance during bigger part of your usage and  scale up to a c4.8xlarge instance whenever required.This flexibility allows you to pay only half of the On-Demand, per-hour price for the larger instance.

For companies with dynamic workloads and architecture, this new flexibility not just reduces the watch one needed to have on the usage of the RI but also opens up the possibility to manage RIs easily at scale without having to do everything manually every time.