Batch Submit API

[egaumer]

In cases where a lot of existing data needs to be indexed, it would be helpful to support batching of documents in an attempt to reduce the message passing overhead associated with HTTP.

The goal would be to submit a batch of documents using a single HTTP call. Tradeoffs associated with such a feature are expected.

[ghost]

Try to use the java-api instead of curl.
On my machine I was able to index 1 Mio documents a 4Kb in 50 minutes :)
Look at that FeederDemo.java:

http://pascal.selfip.org/jforum/posts/list/117.page

Make sure you already have a data-node running (bin/elasticsearch -f should do) and that you execute the FeederDemo.java on the same machine (or at least on a network where the Demo and the data-node see each other).

oo

[kimchy]

Yo, have not forgotten about this issue. Its a bit tricky to implement correctly. Will get 0.7 out of the way and see if I can get back to it.

[kimchy]

Closing as this has been implemented in the bulk api.