Hi community,
Is there any possibility to preload all documents to the filesystem cache using 'index.store.preload' to increase search perfomance of the fetch phase?
Search query is below:
POST /structure/_search?typed_keys=true&max_concurrent_shard_requests=5&ignore_unavailable=false&expand_wildcards=open&allow_no_indices=true&ignore_throttled=true&scroll=60000ms&search_type=dfs_query_then_fetch&batched_reduce_size=512&ccs_minimize_roundtrips=true
{
"from": 0,
"size": 10000,
"version": true,
"explain": false,
"profile": true,
"_source": {
"include": ["externalId"],
"exclude":[]
},
"query": {
"bool": {
"must": [
{
"term": {
"active": {
"value": true,
"boost": 1.0
}
}
},
{
"bool": {
"should": [
{
"terms": {
"path.parent": [
"022D329A-25EA-E711-80FD-001DD8B71DB5"
],
"boost": 1.0
}
},
{
"terms": {
"path.parent": [
"4060741400016833756"
],
"boost": 1.0
}
},
{
"bool": {
"must": [
{
"bool": {
"must": [
{
"terms": {
"data.engagementModel": [
"Managed Capacity"
],
"boost": 1.0
}
},
{
"term": {
"type": {
"value": "STREAM",
"boost": 1.0
}
}
}
],
"adjust_pure_negative": true,
"boost": 1.0
}
},
{
"terms": {
"path.parent": [
"8503000000000008968"
],
"boost": 1.0
}
}
],
"adjust_pure_negative": true,
"boost": 1.0
}
},
{
"terms": {
"path.parent": [
"335FC821-8BD7-E411-80CF-001DD8B71C26",
"6743FA4B-DD01-E811-80FF-001DD8B71DB5",
"C9818B08-206E-EC11-AD3B-00155D00AC2C",
"F2FFA6FA-1F6E-EC11-AD3B-00155D00AC2C",
"32FB7711-860C-EC11-AD33-00155D00AC2C",
"B7534C9C-C643-EC11-AD38-00155D00AC2C",
"34D3E192-A476-E711-80F0-001DD8B71DB5",
"F821EB2A-8ED7-E411-80CF-001DD8B71C26",
"25151EAF-757A-EB11-AD28-00155D00AC29",
"D252FA57-E3F5-EB11-AD32-00155D00AC2C",
"1071F214-16DF-EB11-AD31-00155D00AC2C",
"B3D6D3B5-E298-EA11-AD10-00155D00A803",
"34C3BD1B-8680-E911-811E-001DD8B71DB5",
"3671F1B1-A476-E711-80F0-001DD8B71DB5",
"B61057CC-777A-EB11-AD28-00155D00AC29",
"00E035D3-3263-EC11-AD3C-00155D01722B",
"72A64A3F-767A-EB11-AD28-00155D00AC29",
"370FDA97-DD01-E811-80FF-001DD8B71DB5",
"8E7D5F86-767A-EB11-AD28-00155D00AC29",
"B9116001-9AA2-EB11-AD29-00155D017224",
"123F0385-0CC5-EB11-AD2E-00155D01722B",
"72A14D78-99A2-EB11-AD29-00155D017224",
"06E614E3-98A2-EB11-AD29-00155D017224",
"AA9090EE-97A2-EB11-AD29-00155D017224",
"321758B4-1B53-EA11-AD0C-00155D018201",
"048FC792-AA5A-E911-811C-001DD8B71DB5",
"510EF48E-AB5A-E911-811C-001DD8B71DB5",
"788872ED-AB5A-E911-811C-001DD8B71DB5"
],
"boost": 1.0
}
}
],
"adjust_pure_negative": true,
"boost": 1.0
}
}
],
"adjust_pure_negative": true,
"boost": 1.0
}
}
}
Running result:
-
The first run took 907 ms and fetch phase ~ 700 ms.
-
The second run took 249 ms and fetch phase ~ 212 ms.
As see the second run is mush faster then the first. Is the reason in the filesystem cache and loaded documents to memory or any other reasons?