Handling spikes

[iwan-everlong]

Hi everyone.
Is there a recommended setup to handle spikes?
Our search is simple, our index is small (<1GB).
We have currently 8 replicas and 2 partitions.
But about 2x a day, we have a third party system that hits the search heavly (~30k queries per minute), and we're getting a lot of throttling responses.
We have tried different configurations (6 replicas and 2 partitions, 4 replicas and 4 partitions, etc)

[#2 · Accepted solution]

With as little data you have, I'm not sure adding additional partitions will be helpful to you. I'd recommend scaling replicas up or potentially moving up to a higher SKU (S2's are roughly 4x as powerful as an S1). Check out these performance tips that we recently published: https://docs.microsoft.com/en-us/azure/search/search-performance-tips. There's also some performance benchmarks we recently published: https://docs.microsoft.com/en-us/azure/search/performance-benchmarks. At the end of the day, your best bet will be to run performance tests. We recently published a solution to help bootstrap performance tests for search: https://github.com/Azure-Samples/azure-search-performance-testing.

[#3]

Please do feel free to send me a private message if you'd like to have a more in depth discussion