Cloud cost control: a guide to Reserved Instances


Tech Data Value, Hybrid Cloud
20 Nov 18 Author: Eric Holler

Potential cost savings are driving more and more businesses to the Cloud. Once organisations have deployed a stable environment, they can focus on eliminating waste to optimise cloud spend. What can organisations do to further enhance savings? Achieving effective cost control depends on consolidating all accounts and gaining visibility of cloud usage across the whole business.
Reserved Instances or RIs are a powerful tool in the cost savings toolbox. Introduced by AWS, essentially a method of pre-paying for compute, RIs can provide 30% to 60% in savings over standard pricing. Today, both AWS and Azure offer the ability to purchase RIs.

How Reserved Instances work
The customer promises to use a specific instance type in a specific availability zone all day, every day (8760 hours a year). Reserving the instances in this way, allows the provider to forecast and plan for data centre capacity. In exchange, the RI user receives discounts and priority recovery in a disaster situation. Since inception, there have been a number of changes to the AWS RI program. Locations are more flexible, instance sizes are no longer locked, and in some cases the priority recovery is disabled. The basic premise remains - committed usage in a committed area.

Three things customers need to know before buying an RI

Usage: The customer promises to pay the RI rate all day/every day for the full RI term (8760 hours a year). The challenge is identifying the instances that meet the usage threshold before buying an RI against it. Otherwise, the cloud user could end up paying over the odds.

Geographic Location: When RIs were first released the exact AWS region and availability zone (think data center) for RI deployment had to be identified. Today, RIs can be bought at a regional level. Regional type RIs will automatically deploy in any availability zone within a region.

Instance family, size, and OS: The actual instance type, for example, t2.large running Linux needs to be identified and includes:

  • Size: CPU, memory, storage, and network of the instance. The “large” portion of the above instance is the size. Large in this machine maps to 4 VCPU and 8 GB of ram.
  • OS: The operating system such as AWS Linux, RHEL, or Windows. Discounts vary with different operating systems. RHEL and Windows receive a smaller discount. AWS Linux RIs are a more flexible option inside their families with the latest generation of AWS Linux being dynamically resized by AWS to get the best usage. The t2.large example above will be automatically applied to two t2.mediums or four t2.smalls if a t2.large is not available. Again, this is an AWS Linux only feature and limited to inside the regions.
  • Family: This is the ID of the instance. It correlates to the processor type and role of the instance inside of AWS. The t2 is the family identifier of the above machine.

That simple example goes some way to illustrate the number of variables and potential for complexity. Small wonder that customers often struggle to analysing their current usage and uncover the best opportunities for leveraging RIs. In future blogs, we’ll delve deeper still into the subtleties of RI purchasing and management.

Leverage TD SYNNEX’s comprehensive Cloud and Next-Generation technology portfolio, with the Cloud Solutions team. By gaining access to our commercial and technical enablement, end-to-end services and multi-cloud management, we’re putting you in control.

If you’re an existing partner and want to learn more about TD SYNNEX, or if you’re looking for a valued partner to help accelerate your business, get in contact with your local team today.