The query has several match conditions with fuzziness 2, which makes it quite expensive. I don't want to search for document relevance or any kind of scoring, just identify the list of document IDs that match the query criteria. A simplified version of a query is:
{
"profile": true,
"sort": "_doc",
"query": {
"constant_score": {
"filter": {
"bool": {
"filter": [
{
"bool": {
"should": [
{
"match": {
"namesCombined": {
"query": "name1 name2 name3",
"operator": "AND",
"fuzziness": "2",
"fuzzy_rewrite": "constant_score"
}
}
},
{
"match": {
"namesCombined": {
"query": "name1",
"operator": "AND",
"fuzziness": "2",
"fuzzy_rewrite": "constant_score"
}
}
}
],
"minimum_should_match": "1"
}
}
]
}
},
"boost": 1.0
}
},
"_source": {
"includes": [
"id"
],
"excludes": []
}
}
The above query takes 5ms to execute (when the documents are matched) in an index of 2 documents total, which is a long time for such a simple query. After analyzing with profile API seems that the build_scorer takes 3.8ms. Even though I use bool filter, constant_score queries, and "fuzzy_rewrite":"constant_score", the score calculation slows down the search process. I also noticed that if I use the default rewite method instead of "fuzzy_rewrite":"constant_score" the rewrite_time increases to 3ms and the build_scorer remains low. Is it possible to disable the scoring or improve the performance? Any advice or suggestion would be appreciated.
Profiler results:
{
"profile": {
"shards": [
{
"id": "[v1VNjaOiQrqkDPKg4aTVfA][testindex][0]",
"searches": [
{
"query": [
{
"type": "ConstantScoreQuery",
"description": "ConstantScore(((+namesCombined:NICKOLAS~2 +namesCombined:JACK~2 +namesCombined:DOE~2) namesCombined:JACKDOE~2)~1)",
"time_in_nanos": 3902030,
"breakdown": {
"set_min_competitive_score_count": 0,
"match_count": 0,
"shallow_advance_count": 0,
"set_min_competitive_score": 0,
"next_doc": 6278,
"match": 0,
"next_doc_count": 10,
"score_count": 0,
"compute_max_score_count": 0,
"compute_max_score": 0,
"advance": 9500,
"advance_count": 2,
"score": 0,
"build_scorer_count": 4,
"create_weight": 4852,
"shallow_advance": 0,
"create_weight_count": 1,
"build_scorer": 3881400
},
"children": [
{
"type": "BooleanQuery",
"description": "((+namesCombined:NICKOLAS~2 +namesCombined:JACK~2 +namesCombined:DOE~2) namesCombined:JACKDOE~2)~1",
"time_in_nanos": 3899555,
"breakdown": {
"set_min_competitive_score_count": 0,
"match_count": 0,
"shallow_advance_count": 0,
"set_min_competitive_score": 0,
"next_doc": 5909,
"match": 0,
"next_doc_count": 10,
"score_count": 0,
"compute_max_score_count": 0,
"compute_max_score": 0,
"advance": 9411,
"advance_count": 2,
"score": 0,
"build_scorer_count": 4,
"create_weight": 3493,
"shallow_advance": 0,
"create_weight_count": 1,
"build_scorer": 3880742
},
"children": [
{
"type": "BooleanQuery",
"description": "+namesCombined:NICKOLAS~2 +namesCombined:JACK~2 +namesCombined:DOE~2",
"time_in_nanos": 2702444,
"breakdown": {
"set_min_competitive_score_count": 0,
"match_count": 0,
"shallow_advance_count": 0,
"set_min_competitive_score": 0,
"next_doc": 5518,
"match": 0,
"next_doc_count": 10,
"score_count": 0,
"compute_max_score_count": 0,
"compute_max_score": 0,
"advance": 9329,
"advance_count": 2,
"score": 0,
"build_scorer_count": 6,
"create_weight": 2024,
"shallow_advance": 0,
"create_weight_count": 1,
"build_scorer": 2685573
},
"children": [
{
"type": "MultiTermQueryConstantScoreWrapper",
"description": "namesCombined:NICKOLAS~2",
"time_in_nanos": 1719693,
"breakdown": {
"set_min_competitive_score_count": 0,
"match_count": 0,
"shallow_advance_count": 0,
"set_min_competitive_score": 0,
"next_doc": 0,
"match": 0,
"next_doc_count": 0,
"score_count": 0,
"compute_max_score_count": 0,
"compute_max_score": 0,
"advance": 4243,
"advance_count": 12,
"score": 0,
"build_scorer_count": 6,
"create_weight": 334,
"shallow_advance": 0,
"create_weight_count": 1,
"build_scorer": 1715116
}
},
{
"type": "MultiTermQueryConstantScoreWrapper",
"description": "namesCombined:JACK~2",
"time_in_nanos": 543438,
"breakdown": {
"set_min_competitive_score_count": 0,
"match_count": 0,
"shallow_advance_count": 0,
"set_min_competitive_score": 0,
"next_doc": 810,
"match": 0,
"next_doc_count": 10,
"score_count": 0,
"compute_max_score_count": 0,
"compute_max_score": 0,
"advance": 3695,
"advance_count": 2,
"score": 0,
"build_scorer_count": 6,
"create_weight": 20,
"shallow_advance": 0,
"create_weight_count": 1,
"build_scorer": 538913
}
},
{
"type": "MultiTermQueryConstantScoreWrapper",
"description": "namesCombined:DOE~2",
"time_in_nanos": 416365,
"breakdown": {
"set_min_competitive_score_count": 0,
"match_count": 0,
"shallow_advance_count": 0,
"set_min_competitive_score": 0,
"next_doc": 0,
"match": 0,
"next_doc_count": 0,
"score_count": 0,
"compute_max_score_count": 0,
"compute_max_score": 0,
"advance": 3024,
"advance_count": 12,
"score": 0,
"build_scorer_count": 6,
"create_weight": 20,
"shallow_advance": 0,
"create_weight_count": 1,
"build_scorer": 413321
}
}
]
},
{
"type": "MultiTermQueryConstantScoreWrapper",
"description": "namesCombined:JACKDOE~2",
"time_in_nanos": 1187767,
"breakdown": {
"set_min_competitive_score_count": 0,
"match_count": 0,
"shallow_advance_count": 0,
"set_min_competitive_score": 0,
"next_doc": 0,
"match": 0,
"next_doc_count": 0,
"score_count": 0,
"compute_max_score_count": 0,
"compute_max_score": 0,
"advance": 0,
"advance_count": 0,
"score": 0,
"build_scorer_count": 2,
"create_weight": 19,
"shallow_advance": 0,
"create_weight_count": 1,
"build_scorer": 1187748
}
}
]
}
]
}
],
"rewrite_time": 28163,
"collector": [
{
"name": "SimpleFieldCollector",
"reason": "search_top_hits",
"time_in_nanos": 75426
}
]
}
],
"aggregations": []
}
]
}
}
A sample document:
{"id":"id","namesCombined":"name1 name2 name3"}
Index settings and mapping:
{
"settings": {
"index": {
"number_of_shards": 1,
"number_of_replicas": 0,
"store.type": "mmapfs",
"refresh_interval": "300s"
}
},
"mappings": {
"namesCombined": {
"type": "text",
"analyzer": "whitespace",
"doc_values": false,
"index_options": "docs",
"norms": false,
"store": false,
"term_vector": "no"
}
}
}