Using ElasticSearch as a storage for classifiers


I'm working on application that should store classifiers in the ES with searching capability. Some classifiers are simple set of key-value pars, some have a complex structure. I've decided to keep each classifier's data in separate index. My application have about 1000 classifiers and when I am trying to import them from external sources, I get troubles with elasticsearch (it crushes with not connected answer on my local PC). I use all default settings during indexing (5 shards per index, 1 node on my local PC).
I think the issue is related to huge amount of shards which getting created during import processing.

Can you point on possible reasons of this situation and which indexing strategy can be used for my situation?