Performance of ExecutionStrategy#completeField and ExecutionStrategy#fetchField

Hi,

While testing our application for 1k data, in the JProfiler readings we see graphql parsing and processing is taking significantly more time than the DB layer.

Specifically two methods:

• graphql.execution.ExecutionStrategy#completeField
• graphql.execution.ExecutionStrategy#fetchField

*marked in blue is the Datafetcher implementation that looks up data from DB, taking 13.4% of the total execution time Note: the output of DataFetchers is Map<String, Object>

• Query used for this call:

{
  "query": "query searchEntities($scope: SearchScope!,$first: Int!,$sortBy: SortByField) {searchEntities(scope: $scope,first:$first,sortBy: $sortBy) {matchedEntities { activeEntity {  name description } draftEntities{ description probableEntity{ description name createTime }  }} pageInfo { endCursor hasNextPage } } } ",
  "variables": {    
	"scope": {
      "appIdentifier": "app1",
      "nativeType": "nativeType1"
    },
    "first":1000,
    "sortBy":{
    	"field":"UpdateTime",
    	"order":"Desc"
    }
  }
}

We are using graphql-java version : 14.0. Also attaching graphql schema. EntityApi.graphql.txt

---

• looking up 65k records, total Graphql execution took about 228s

  where actual DB lookup took ~22s only

Question is : What factors contribute to graphql processing overhead? Are there any best practices that can help to reducing the overall time as we are extensively using graphql in all our APIs?

[bbakerman]

I am going t close this issue. We have attackd the performance of the execution strategy quite a lot of late.

Any of the (very welcome) perf details above will not be out of date to the code.

We should raise new performance issues against the latest code at future points

[bbakerman]

we’re using graphql java kickstart

In which case I dont know how kickstart + kotlin works and this project (graphql-java) does not support kick start. Kick start is built on top of graphql-java but how it uses it exactly is not something I can help with.

Others may know more.

Good luck with finding your answers