I have an index an cluster on Amazon Opensearch Service.
My index is about 13gb and the configuration is 2 shards and 1 replica.
When I re-indexed using the Reindex API the whole migration took about 52 min.
Is this time normal or can I reduce it by performing some configuration?
CPU = 2
Memory = 8gb
EBS = 100gb
Master instance = 3
Data nodes = 2
The time spent is normal, by default reindex will use one single thread to scroll the source index and then indexing to the target index, the default batch size is 1000 when scrolling. You can improve the performance by increasing the batch size and increasing parallelism(use slicing), like this:
In the next indexation I will use the mentioned parameters. thanks