Home  /  Installation

Operational guidelines

  • ARM-based CPU
  • 1:8 ratio of vCPU to GiB memory
  • 1:16 ratio of vCPU to GiB local instance storage (if enabling spill-to-disk)

When operating in AWS, we recommend:

  • Using the r7gd and r6gd families of instances (and r8gd once available) when running with local disk

  • Using the r8g, r7g, and r6g families when running without local disk

CPU affinity

It is strongly recommended to enable the Kubernetes static CPU management policy. This ensures that each worker thread of Materialize is given exclusively access to a vCPU. Our benchmarks have shown this to substantially improve the performance of compute-bound workloads.

Locally-attached NVMe storage

For optimal performance, Materialize requires fast, locally-attached NVMe storage. Having a locally-attached storage allows Materialize to spill to disk when operating on datasets larger than main memory as well as allows for a more graceful degradation rather than OOMing. Network-attached storage (like EBS volumes) can significantly degrade performance and is not supported.

Additional documentation to come

See also

Back to top ↑