Has anyone tried running Elasticsearch on an X1 instance with 2Tb of RAM?

AWS just launched its X1 instance: https://aws.amazon.com/ec2/instance-types/x1/

X1 Instances are a new addition to the Amazon EC2 memory-optimized instance family and are designed for running large-scale, in-memory applications and in-memory databases in the AWS cloud. X1 instances offer 1,952 GiB of DDR4 based memory, 8x the memory offered by any other Amazon EC2 instance. Each X1 instance is powered by four Intel® Xeon® E7 8880 v3 (Haswell) processors and offers 128 vCPUs.

Has anyone tried running Elasticsearch on one of these instances? Would it make sense to try it?

No, but never let that stop you :stuck_out_tongue:

Could you elaborate? I know that Elasticsearch is designed to be massively horizontally scalable and that you incur a cost by going with more than 32Gb of RAM because addresses become longer.
That said, I would imagine that you could improve performance on huge clusters by leveraging bigger hosts to save on network costs. Is that not the case?

You incur cost by having more than 32gb of heap. But 2tb of filesystem
cache is still useful. So long as you have a large enough working set.

1 Like