I'm about to use Rally to benchmark a cluster on elastic.co so that we can get a better idea of our needs. I'd like to use an AWS EC2 instance for Rally so that we don't have the office network as a bottle neck. What sort of specs should I use for the server specs? I couldn't find anything in the documentation other than to use an SSD.
the docs mention SSDs as this is crucial to avoid a potential client-side bottleneck due to Rally's disk access pattern. Apart from that it really depends on your specific benchmarking scenario. One important factor is definitely the network connection between Rally and Elasticsearch. The load generator should be close to the target machines and you should have a fast connection between them; it is possible that a 1GBit connection is not sufficient but you need to measure this. The blog post Seven Tips for Better Elasticsearch Benchmarks contains more specific tips.
This topic was automatically closed 28 days after the last reply. New replies are no longer allowed.