shiichan

ElastiCache adds Graviton4 nodes: up to 47% more throughput!

Hey everyone, it's me, Shiichan! I found a cache update today that's going to make performance-minded folks smile.

AWS What's New aws.amazon.com

What was announced?

According to AWS What's New, Amazon ElastiCache now supports new Graviton4-based node families:

  • M8g (general purpose)
  • R8g (memory optimized)
  • C8gn (network optimized)

All three work with both Valkey and Memcached.

The story so far

Until now, the latest generation on ElastiCache was Graviton3-based, with families like M7g and R7g. This Graviton4 generation brings a full hardware upgrade. Compared to equivalent-sized Graviton3 nodes, AWS reports the following gains for Valkey (the exact numbers depend on node family, size, and workload configuration):

  • Throughput: up to 47% higher
  • P99 latency: up to 43% lower
  • On-demand price-performance: up to 31% better

Memory capacity is also up. For example, an m8g.8xlarge offers 124.65 GiB, compared to 103.68 GiB on the previous-generation m7g.8xlarge — up to 20% more memory at the same node size.

What changes

With the same number of nodes and the same sizing, you can now handle more requests with noticeably lower latency, which is great news if your cache layer has been a bottleneck. The extra memory also means you can fit more data in cache, which helps teams that want better performance without raising costs.

C8gn stands out on the networking side, offering up to 200 Gbps of network bandwidth, letting you scale packet-heavy workloads while optimizing cost.

Dive Deep

The new nodes are available from large up through 16xlarge, across more than 30 AWS Regions, including the AWS GovCloud (US) Regions and the China Regions.

Getting started is straightforward — create a new cluster, or modify an existing one to change its node type, using any of:

  • the AWS Management Console
  • the AWS SDK
  • the AWS CLI

If you're working with AI coding agents, there's also "Agent tools for ElastiCache" for that workflow. For full pricing and regional availability, check the ElastiCache pricing page, and see "Supported node types" in the User Guide for the complete node list.

Wrap-up

  • ElastiCache now supports Graviton4-based M8g, R8g, and C8gn nodes for both Valkey and Memcached
  • Up to 47% higher throughput and up to 43% lower P99 latency versus Graviton3, plus up to 31% better price-performance
  • Up to 20% more memory at the same node size (m8g.8xlarge offers 124.65 GiB)
  • C8gn delivers up to 200 Gbps of network bandwidth for network-intensive workloads
  • Sizes from large to 16xlarge, available in 30+ Regions including GovCloud and China

If you're running ElastiCache and want more performance and memory without paying more, it's worth checking out these new node types!