Better performance for Bulk API

This commit has a few performance-related changes. It comes in two flavours. First, the way that JSON got serialized was changed from using a simple `map[string]interface{}` to a specific Go type. That had the following impact: ``` benchmark old ns/op new ns/op delta BenchmarkBulkIndexRequestSerialization-8 3348 2285 -31.75% benchmark old allocs new allocs delta BenchmarkBulkIndexRequestSerialization-8 34 18 -47.06% benchmark old bytes new bytes delta BenchmarkBulkIndexRequestSerialization-8 2464 1744 -29.22% benchmark old ns/op new ns/op delta BenchmarkBulkUpdateRequestSerialization-8 3131 2091 -33.22% benchmark old allocs new allocs delta BenchmarkBulkUpdateRequestSerialization-8 39 19 -51.28% benchmark old bytes new bytes delta BenchmarkBulkUpdateRequestSerialization-8 2624 2624 +0.00% benchmark old ns/op new ns/op delta BenchmarkBulkDeleteRequestSerialization-8 2233 1158 -48.14% benchmark old allocs new allocs delta BenchmarkBulkDeleteRequestSerialization-8 28 11 -60.71% benchmark old bytes new bytes delta BenchmarkBulkDeleteRequestSerialization-8 1744 1664 -4.59% ``` Next, we enabled a setting that makes Bulk API use `github.com/mailru/easyjson` instead of `encoding/json`. This yields the following results (compared to the baseline): ``` benchmark old ns/op new ns/op delta BenchmarkBulkIndexRequestSerialization-8 3348 1692 -49.46% benchmark old allocs new allocs delta BenchmarkBulkIndexRequestSerialization-8 34 12 -64.71% benchmark old bytes new bytes delta BenchmarkBulkIndexRequestSerialization-8 2464 1328 -46.10% benchmark old ns/op new ns/op delta BenchmarkBulkUpdateRequestSerialization-8 3131 1072 -65.76% benchmark old allocs new allocs delta BenchmarkBulkUpdateRequestSerialization-8 39 10 -74.36% benchmark old bytes new bytes delta BenchmarkBulkUpdateRequestSerialization-8 2624 1952 -25.61% benchmark old ns/op new ns/op delta BenchmarkBulkDeleteRequestSerialization-8 2233 593 -73.44% benchmark old allocs new allocs delta BenchmarkBulkDeleteRequestSerialization-8 28 5 -82.14% benchmark old bytes new bytes delta BenchmarkBulkDeleteRequestSerialization-8 1744 1280 -26.61% ``` You can enable serialization with `UseEasyJSON(true)` on the individual `BulkIndexRequest`, `BulkUpdateRequest`, and `BulkDeleteRequest`. Using easyjson is **experimental** for now. As easyjson relies on code generation, you need to use `go generate` when structs change, in order to create the underlying structs that easyjson makes use of. See #667 for discussion.
olivere · Jan 12, 2018 · e771061 · e771061
1 parent d2219c2
commit e771061
Show file tree

Hide file tree

Showing 11 changed files with 1,194 additions and 152 deletions.
diff --git a/bulk_delete_request.go b/bulk_delete_request.go
@@ -4,6 +4,8 @@
 
 package elastic
 
+//go:generate easyjson bulk_delete_request.go
+
 import (
 	"encoding/json"
 	"fmt"
@@ -27,13 +29,39 @@ type BulkDeleteRequest struct {
 	versionType string // default is "internal"
 
 	source []string
+
+	useEasyJSON bool
+}
+
+//easyjson:json
+type bulkDeleteRequestCommand map[string]bulkDeleteRequestCommandOp
+
+//easyjson:json
+type bulkDeleteRequestCommandOp struct {
+	Index       string `json:"_index,omitempty"`
+	Type        string `json:"_type,omitempty"`
+	Id          string `json:"_id,omitempty"`
+	Parent      string `json:"_parent,omitempty"`
+	Routing     string `json:"_routing,omitempty"`
+	Version     int64  `json:"_version,omitempty"`
+	VersionType string `json:"_version_type,omitempty"`
 }
 
 // NewBulkDeleteRequest returns a new BulkDeleteRequest.
 func NewBulkDeleteRequest() *BulkDeleteRequest {
 	return &BulkDeleteRequest{}
 }
 
+// UseEasyJSON is an experimental setting that enables serialization
+// with github.com/mailru/easyjson, which should in faster serialization
+// time and less allocations, but removed compatibility with encoding/json,
+// usage of unsafe etc. See https://github.com/mailru/easyjson#issues-notes-and-limitations
+// for details. This setting is disabled by default.
+func (r *BulkDeleteRequest) UseEasyJSON(enable bool) *BulkDeleteRequest {
+	r.useEasyJSON = enable
+	return r
+}
+
 // Index specifies the Elasticsearch index to use for this delete request.
 // If unspecified, the index set on the BulkService will be used.
 func (r *BulkDeleteRequest) Index(index string) *BulkDeleteRequest {
@@ -106,39 +134,32 @@ func (r *BulkDeleteRequest) Source() ([]string, error) {
 	if r.source != nil {
 		return r.source, nil
 	}
-	lines := make([]string, 1)
-
-	source := make(map[string]interface{})
-	deleteCommand := make(map[string]interface{})
-	if r.index != "" {
-		deleteCommand["_index"] = r.index
+	command := bulkDeleteRequestCommand{
+		"delete": bulkDeleteRequestCommandOp{
+			Index:       r.index,
+			Type:        r.typ,
+			Id:          r.id,
+			Routing:     r.routing,
+			Parent:      r.parent,
+			Version:     r.version,
+			VersionType: r.versionType,
+		},
 	}
-	if r.typ != "" {
-		deleteCommand["_type"] = r.typ
-	}
-	if r.id != "" {
-		deleteCommand["_id"] = r.id
-	}
-	if r.parent != "" {
-		deleteCommand["_parent"] = r.parent
-	}
-	if r.routing != "" {
-		deleteCommand["_routing"] = r.routing
-	}
-	if r.version > 0 {
-		deleteCommand["_version"] = r.version
-	}
-	if r.versionType != "" {
-		deleteCommand["_version_type"] = r.versionType
-	}
-	source["delete"] = deleteCommand
 
-	body, err := json.Marshal(source)
+	var err error
+	var body []byte
+	if r.useEasyJSON {
+		// easyjson
+		body, err = command.MarshalJSON()
+	} else {
+		// encoding/json
+		body, err = json.Marshal(command)
+	}
 	if err != nil {
 		return nil, err
 	}
 
-	lines[0] = string(body)
+	lines := []string{string(body)}
 	r.source = lines
 
 	return lines, nil

diff --git a/bulk_delete_request_easyjson.go b/bulk_delete_request_easyjson.go
diff --git a/bulk_delete_request_test.go b/bulk_delete_request_test.go
@@ -17,21 +17,21 @@ func TestBulkDeleteRequestSerialization(t *testing.T) {
 		{
 			Request: NewBulkDeleteRequest().Index("index1").Type("doc").Id("1"),
 			Expected: []string{
-				`{"delete":{"_id":"1","_index":"index1","_type":"doc"}}`,
+				`{"delete":{"_index":"index1","_type":"doc","_id":"1"}}`,
 			},
 		},
 		// #1
 		{
 			Request: NewBulkDeleteRequest().Index("index1").Type("doc").Id("1").Parent("2"),
 			Expected: []string{
-				`{"delete":{"_id":"1","_index":"index1","_parent":"2","_type":"doc"}}`,
+				`{"delete":{"_index":"index1","_type":"doc","_id":"1","_parent":"2"}}`,
 			},
 		},
 		// #2
 		{
 			Request: NewBulkDeleteRequest().Index("index1").Type("doc").Id("1").Routing("3"),
 			Expected: []string{
-				`{"delete":{"_id":"1","_index":"index1","_routing":"3","_type":"doc"}}`,
+				`{"delete":{"_index":"index1","_type":"doc","_id":"1","_routing":"3"}}`,
 			},
 		},
 	}
@@ -58,11 +58,22 @@ func TestBulkDeleteRequestSerialization(t *testing.T) {
 var bulkDeleteRequestSerializationResult string
 
 func BenchmarkBulkDeleteRequestSerialization(b *testing.B) {
-	r := NewBulkDeleteRequest().Index(testIndexName).Type("doc").Id("1")
+	b.Run("stdlib", func(b *testing.B) {
+		r := NewBulkDeleteRequest().Index(testIndexName).Type("doc").Id("1")
+		benchmarkBulkDeleteRequestSerialization(b, r.UseEasyJSON(false))
+	})
+	b.Run("easyjson", func(b *testing.B) {
+		r := NewBulkDeleteRequest().Index(testIndexName).Type("doc").Id("1")
+		benchmarkBulkDeleteRequestSerialization(b, r.UseEasyJSON(true))
+	})
+}
+
+func benchmarkBulkDeleteRequestSerialization(b *testing.B, r *BulkDeleteRequest) {
 	var s string
 	for n := 0; n < b.N; n++ {
 		s = r.String()
 		r.source = nil // Don't let caching spoil the benchmark
 	}
 	bulkDeleteRequestSerializationResult = s // ensure the compiler doesn't optimize
+	b.ReportAllocs()
 }