AWS has slashed prices on its container friendly Fargate service, leaving EC2 users with some sums to do as to where best to run their workloads.
As AWS puts it, “AWS Fargate is a compute engine that uses containers as its fundamental compute primitive [and] runs your application containers for you on demand.”
The idea is to remove provisioning, and having to manage an agent or daemon, making the underlying infrastructure “invisible” and avoid paying for resources you don’t use. Yes, it’s like serverless, but without having to do so much rearchitecting of your legacy apps.
To date, the service has not been cheap, leaving little incentive for many users to shift off their EC2 instances.
However, AWS says that the introduction of its Firecracker microVM tech late last year means that as of this week, it can afford to slash pricing per vCPU by 20 per cent and pricing per GB of memory per second by 65 per cent.
That works out to a reduction of 35 per cent to 50 per cent, depending on configuration. You can see the various permutations here.
More to the point, AWS says “If your application is currently running on large EC2 instances that peak at 10-20% CPU utilization, consider migrating to containers in AWS Fargate.”
One AWS partner, Trek10, calculated that “break-even between Fargate and EC2 now happens in the 60-80% reservation rate.” So, if you’ve got a cluster that is only 50 per cent utilized, moving to Fargate could save you 10 to 20 per cent. If you’ve got very high utilization rate, moving to Fargate could increase costs, says Trek10, but companies should consider the management overhead this degree of utilization demands.