I am guessing that is purely metadata like mentioned in the documentation and doesn't occupy too much space. We have observed that having too many indices/shards slows down the cluster and there isn't any easy way to recovery from that state. For example, we have 47k shards on a cluster of 8 nodes (5 data + 3 dedicated master nodes).

We also have dedicated master nodes (3 to be specific), so suppose the that master nodes take care keeping index information.

---
NEW: Monitor These Apps!
elasticsearch, apache solr, apache hbase, hadoop, redis, casssandra, amazon cloudwatch, mysql, memcached, apache kafka, apache zookeeper, apache storm, ubuntu, centOS, red hat, debian, puppet labs, java, senseiDB