Idle way of creating indeces


(avinash.nalluri) #1

Hi,
Can anyone please tell me what is an idle number of shards to be
created in a particular index for large data operations and how does
those affects querying mechanisms(speed) also,considering the fact
that we may scale our elastic search cluster from single machine what
we are using now to multiple(more than 3) machines.

Thanks,
avinash


(Karussell) #2

That really depends on your data. Querying can be potentially faster
as three machines can work on the same query but I doubt this is the
case as every machine has at least 3 core nowadays and network is the
most limiting resource. I would try (hard) to split the index so that
you can dynamically add and remove indices instead of sharding. If
this isn't an option changing the shards is just a re-index away :wink:
(relative easy via scan search)

Peter.

On Nov 8, 5:15 pm, "avinash.nalluri" avinash.nall...@gmail.com
wrote:

Hi,
Can anyone please tell me what is an idle number of shards to be
created in a particular index for large data operations and how does
those affects querying mechanisms(speed) also,considering the fact
that we may scale our elastic search cluster from single machine what
we are using now to multiple(more than 3) machines.

Thanks,
avinash


(system) #3