Hello,
we are trying out the elastic search (version 5.5) percolate for our notification feature. But we see the performance is really slow.
{"query": {
"bool": {
"must": [
{
"bool": {
"should": [
{
"term": {
"cType": {
"value": "R",
"boost": 1
}
}
}
],
"disable_coord": false,
"adjust_pure_negative": true,
"minimum_should_match": "1",
"boost": 1
}
},
{
"terms": {
"oType": [
"O"
],
"boost": 1
}
},
{
"geo_shape": {
"actualLocation": {
"shape": {
"type": "polygon",
"orientation": "right",
"coordinates": [
[
[
[
1.3313,
4.0267
],
[
1.3848,
4.0302
],
[
1.4362,
4.0356
],
[
1.4836,
4.0521
],
[
1.5252,
4.0744
],
[
1.5595,
4.1016
],
[
1.5802,
4.1326
],
[
1.596,
4.1662
],
[
1.6015,
4.201
],
[
1.5964,
4.2362
],
[
1.5809,
4.2699
],
[
1.5556,
4.301
],
[
1.5214,
4.3283
],
[
1.4797,
4.3507
],
[
1.4321,
4.3674
],
[
1.3803,
4.3777
],
[
1.3301,
4.3809
],
[
1.2823,
4.3778
],
[
1.2305,
4.3675
],
[
1.1829,
4.3509
],
[
1.1412,
4.3285
],
[
1.107,
4.3012
],
[
1.0817,
4.2701
],
[
1.0662,
4.2363
],
[
1.061,
4.2013
],
[
1.0666,
4.1662
],
[
1.0824,
4.1326
],
[
1.1031,
4.1016
],
[
1.1374,
4.0744
],
[
1.179,
4.0521
],
[
1.2264,
4.0356
],
[
1.2779,
4.0302
],
[
1.3313,
4.0267
]
]
]
]
},
"relation": "within"
},
"ignore_unmapped": false,
"boost": 1
}
}
],
"disable_coord": false,
"adjust_pure_negative": true,
"boost": 1
}
}
}
We tried to use 50000 such queries and tried to percolate 1 document. All the queries will have a geo-shape as we are dealing with location based documents.
Questions.
we have 50000 queries with geo shape in it. In spite of filtering with other criteria, I believe elastic search should at least check the geo-shape for 10000 - 20000 queries.
- Is the performance expected ?
- Are you seeing any changes in our query that will increase the performance drastically ?
Kindly help.
Thank you,
Pradeep.