Why do multiple nested objects appear in this ElasticSearch query? - database

Question

In Case 4 of this page, the query searches for all chairs less than 70 units in height:
curl localhost:9200/example/product/_search -d '{
"query": {
"filtered": {
"query": {
"match": {
"name": "chair"
}
},
"filter": {
"numeric_range": {
"size.height": {
"lt": 70
}
}
}
}
}
}'
Result:
"hits": [
{
"_id": "0",
"_source": {
"product": "chair",
"size": [
{
"width": 50,
"height": 50,
"depth": 50
},
{
"width": 75,
"height": 75,
"depth": 75
}
]
}
}
]
1) why is the ID 0 for both chair sizes?
2) why does the response show dimensions for the other chair that is 75 units in height?

Answer 1

1) The writer wanted to show 1 to N relation. Meaning there are 2 (In this case) types of chairs in his repository: A chair with dimensions of 50 and a chair with dimensions of 75. But both of them are still chairs and the id of a chair is 0.
2) Because by default ES doesn't return partial results, it returns documents. In our case we have a chair document with a size array which holds 2 objects: One for the 50 dimension and one for the 75 dimension. The supplied query can either select the whole document or not.
If you want to convert the query to English you may say: Bring me all the documents which have the value "chair" in the product field and at least one of its size.height values is lower than 70.
Even though the writer of the article is knowledgeable, I must say I don't like this kind of articles that trying to draw a direct flow between the SQL world to NOSQL implementation. If it was so easy, some big company would have write an automatic script that converts your SQL schemas to various NOSQL formats. In order to model your data correctly in NOSQL you must understand your products, understand the factors that should influence on your decision, understand the use case and the data. There is no one universal solution that will tell you: If you did it this way in a RDBMS do it like this in ES.