Setting timezone on date_histogram slow down the query

[Bertrand (Bertrand Renuart)]

Hi there,

It seems that setting the timezone option to a date_histogram has a negative effect on the query performance. Suppose the following query:

{
  "size": 0,
  "aggs": {
    "2": {
      "date_histogram": {
        "field": "@timestamp",
        "calendar_interval": "1d",
        "time_zone": "Europe/Brussels"
      }
    }
  },
  "query": {
    "bool": {
      "must": [
        {
          "range": {
            "@timestamp": {
              "format": "strict_date_optional_time",
              "gte": "2019-05-01T00:00:00.000Z",
              "lte": "2019-05-10T00:00:00.000Z"
            }
          }
        }
      ]
    }
  }
}

This query is executed on an alias targeting 10 indexes, each with 3 shards and 1 replica - all hosted on 3 different nodes and containing a total of 480M documents.

With timezone set, execution time is 16821ms. Without timezone, the execution time goes down to 6920ms.

Is this expected?
Is there something I can do to optimize the setup and lower the execution time when timezone is set?

PS: this query is similar to what the Kibana Discover panel does when building the histogram...