Running OpenSearch term aggregations in parallel

Question

We have a query calculating number of terms on multiple fields.

{
 "query": {
 "bool": {
 "filter": [
 {
 "term": {
 "your_filter_field": "your_filter_value"
 }
 }
 ]
 }
 },
 "aggs": {
 "aggregation_1": {
 "terms": {
 "field": "field_1"
 }
 },
 "aggregation_2": {
 "terms": {
 "field": "field_2"
 }
 },
 ...
 "aggregation_50": {
 "terms": {
 "field": "field_3"
 }
 }
 }
}

What we observe is that the aggregations are executed serially (OpenSearch 2.17). As a workaround we use multisearch sending each aggregation in its own query, which is faster, but it feels inefficient as we are executing the same filter multiple times. Is it possible to send multiple aggregations in one query while executing them in parallel?

• Musab Dogan: Are you looking for _msearch?. opensearch.org/docs/latest/api-reference/multi-search
• Lukáš Křečan: We use _msearch as workaround, but it executes the filter for each aggregation. When the filter is expensive, it's waste of resources.
• Musab Dogan: Not sure what is your main pain point but here is a similar discussion. > If you send 4 separate aggregation in parallel over in an _msearch query they run in parallel across all shards. discuss.elastic.co/t/does-aggregations-runs-in-parallel/237189/…