ElastiCache and OpenSearch are managed data services on AWS that bill primarily for the nodes you run and, for OpenSearch, the storage attached to them. They tend to drift expensive because clusters are provisioned for a worst case peak, sized generously for safety, and then never revisited. The four levers that move the bill are node rightsizing to actual utilization, storage tiering for OpenSearch so old data leaves fast disks, migration to Graviton node types for better price for performance, and reserved nodes to discount the steady baseline. Apply them in that order, because committing to oversized nodes just locks in waste.
These are common quiet line items on a mature AWS bill. Here is how to bring each under control.
How do you rightsize ElastiCache nodes?
ElastiCache for Redis or Memcached bills per node hour by node type. The usual waste is a cluster sized for a peak that rarely arrives, running large nodes at low memory and CPU utilization around the clock. Look at sustained memory usage, evictions, and CPU across the cluster, then move to a smaller node type or fewer nodes where the headroom is excessive. Right size first, because the reserved node discount you buy later applies to whatever size you commit to, so committing to a bloated cluster bakes the waste into a one or three year term.
The second lever is Graviton. ElastiCache offers Graviton based node types that deliver better price for performance than the equivalent x86 nodes, so migrating the cluster is a standing win that needs no ongoing management once done.
What drives OpenSearch cost, and how do you tier it?
OpenSearch Service cost is driven by data node hours and the volume of storage attached to them. Keeping every index on hot data nodes with fast storage is the expensive default. The fix is tiering: move older, rarely queried indices to UltraWarm and then to cold storage, which hold the same data at a fraction of the hot tier cost while keeping it searchable. Index lifecycle policies automate the transition so data ages down the tiers without manual effort.
Alongside tiering, rightsize the data nodes to the query and indexing load they actually carry, and use Graviton based instance types where available. Together, tiering and rightsizing usually move the OpenSearch bill more than any single commitment.
When should you buy reserved nodes?
Both services offer reserved nodes: you commit to a node type for one or three years and receive a discount of roughly 30 to 60 percent on the node hourly rate, which is indicative and varies by node type and term. Reserved nodes are a baseline instrument, like Reserved Instances for databases. Reserve only the node capacity you are confident you will run for the term, after rightsizing, and leave the variable top layer on demand.
The sequence matters. Rightsize and tier first so you know your true steady baseline, then reserve that baseline. Buying reservations before rightsizing locks you into paying for nodes you could have shrunk.
A worked example
A scaling fintech ran an OpenSearch cluster that kept a year of logs on hot data nodes and an ElastiCache fleet sized for a launch peak that had long passed. Setting index lifecycle policies to move logs older than 30 days to UltraWarm and then cold, rightsizing both clusters to real load, and migrating to Graviton node types cut the combined node and storage bill substantially. Only after rightsizing did they reserve the steady baseline. Figures are verified against billing data and anonymised, and the discount ranges are indicative.
Frequently asked questions
How do reserved nodes work for ElastiCache and OpenSearch?
What drives OpenSearch cost most?
Can Graviton cut ElastiCache and OpenSearch cost?
Tighten your data service spend with us
We rightsize ElastiCache and OpenSearch to real load, set the storage tiering and lifecycle policies that take data off hot disks, and time reserved node purchases so you commit only to a clean baseline. Our guarantee: we reduce your cloud spend or we reimburse our service fee. Pricing is either a Fixed Fee scoped up front or Gainshare, a share of verified savings with no retainer and no risk.
Put a defensible number on your cloud spend.
No provider in the room, no published price list. Tell us your footprint and we will scope the savings against your billing data — we reduce your cloud spend or we reimburse our service fee.
The Cloud Spend Navigator: what changed in cloud pricing, commitments, and FinOps — no vendor spin.