i'm doing a test on bulk-index.
first, i indexed 1million docs into a single machine, with the bulk
size equals to 5000. the process finished in less than 6min, which is
second, i moved to a cluster environment (5 machines, including the
machine in the first test), and the same bulk size, but the speed is
much slower, which runs 12.5min.
is there a way to make the index any faster? (at least the same as the
single machine test). as i plan to load 200million docs into a 10
machine cluster. the speed would become unacceptable if the index
takes to long.
ps, i have routing set in every single request.
any suggestions and advice would be appreciated, thanks!