bulk.md

Bulk Operations

Elasticsearch allows us to index, delete and update in bulk mode for much faster throughput. When using the bulk API you will save at the very least the latency of multiple requests, and usually Elasticsearch can optimize when it knows it will be doing multiple requests in the same index.

Elastic4s supports bulk operations in an easy way. Index, update and delete operations use the same syntax as before except they are now wrapped in a bulk keyword.

For example, bulk index requests:

val resp = client.execute {
  bulk(
    indexInto("bands").fields("name" -> "coldplay"),
    indexInto("bands").fields("name" -> "kings of leon"),
    indexInto("bands").fields("name" -> "elton john", "best_album" -> "tumbleweed connection")
  )
}

An example of bulk delete operations:

val resp = client.execute {
  bulk(
    deleteById("places", "3"),
    deleteById("artists", "2"),
    deleteById("places", "4"),
  )
}

The bulk API supports combining different type of operations: updates, inserts, deletes.

val resp = client.execute {
  bulk(
    indexInto("bands").fields("name" -> "coldplay"),
    indexInto("bands").fields("name" -> "kings of leon"),
    deleteById("places", "3"),
    deleteById("artists", "2"),
    updateById("bands", "4").doc("name" -> "kate bush")
  )
}

Responses

The response object contains how many successes and failures.

val f = client.execute {
  bulk(
    indexInto("bands").fields("name" -> "coldplay"),
    indexInto("bands").fields("name" -> "kings of leon"),
    deleteById("places", "3"),
    deleteById("artists", "2"),
    updateById("bands", "4").doc("name" -> "kate bush")
  )
}
f.map { resp =>
  println("Failures: ${resp.failures}")
  println("Successes: ${resp.successes}")
}