But, if you want to index a lot of small fields, you can probably use bigger machines. In my use case, I have to index 2 very big fields, so they use one core each and takes like an hour at every restart or problem (and gives problems to sync in the cluster because it's too big)