AWS ElasticCache Serverless – Is it cheaper?

In re:Invent 2023 AWS announced the new ElasticCache serverless, and one of the claims in the announcement release is “You can create a highly available cache with data automatically replicated across multiple Availability Zones and up to 99.99 percent availability Service Level Agreement (SLA) for all workloads, which saves you time and money.”   To test that assertion, I used the AWS calculator to evaluate the cost of a cluster that will have a 100GB cache, with 1 request per second to access the cache. Then I checked what cluster would be of a similar cost, and the result was unsettling: A m6g.16xlarge 3 node cluster providing 209GB of cache, with 100% utilisation (meaning more than 1 request per second) paid in On-Demand prices. Going for a comparison based on the size of the memory cache, I used a 3-node cluster of m8g.8xlarge: The price difference is astounding and does not show a saving for the user. To make the comparison suitable to both type of usage, I’ll now test the calculation for a small cache: 512MB (the minimal size) Serverless Standard cluster The machines with the same memory allocations are the t4g.micro, so a 3-node cluster in OnDemand prices. The “Classic” cluster is still $10.59 cheaper than the same cache in the serverless model.