Following the documentation's canonical example, say I want to store tweets in elasticsearch, but I want to specify an endpoint that handles more than two fields, e.g.
/tweets/some_twitter_handle/some_month/tweet/id
In this example, I want to index on tweets, but also on some_twitter_handle and some_month, with tweet type, and some id.
Is this possible, or is the only possible mapping something akin to what they provide, with one index and one type, e.g /tweets/tweet/id?
The index/type/id is what determines which shard a document ends up on (in the absense of other routing information). Elasticsearch indexes every field by default (you can control this through a mapping if you don't want some of your field indexed). So by putting a "twitter_handle": "whatever" into the document, it becomes searchable by that "twitter_handle".
/tweets/tweet/id (where tweet would house a twitter_handle) more computationally expensive than having the mapping: /tweets/twitter_handle/id? That's mostly my concern.
/tweets/tweet/id would be a specific twitter message. If you were to try to put all of a specific person's tweets into a single document (ie if id were a twitter handle), it would be very inefficient./tweets/twitter_handle/id would still be a single twitter message, but in the former example, the twitter_handle would be a field in the tweet, whereas in the latter, it's just a more specific routing.