perf: optimize deduplication

api/ingester/v1/ingester.proto

@@ -136,8 +136,10 @@ message MergeSpanProfileResult {
 }
 
 message ProfileSets {
+  // DEPRECATED: Use fingerprints instead.
   repeated types.v1.Labels labelsSets = 1;
   repeated SeriesProfile profiles = 2;
+  repeated uint64 fingerprints = 3;
 }
 
 message SeriesProfile {

api/openapiv2/gen/phlare.swagger.json

@@ -1071,14 +1071,22 @@
           "items": {
             "type": "object",
             "$ref": "#/definitions/v1Labels"
-          }
+          },
+          "description": "DEPRECATED: Use fingerprints instead."
         },
         "profiles": {
           "type": "array",
           "items": {
             "type": "object",
             "$ref": "#/definitions/v1SeriesProfile"
           }
+        },
+        "fingerprints": {
+          "type": "array",
+          "items": {
+            "type": "string",
+            "format": "uint64"
+          }
         }
       }
     },

pkg/phlaredb/block_querier.go

@@ -49,7 +49,7 @@ import (
 )
 
 const (
-	defaultBatchSize = 4096
+	defaultBatchSize = 64 << 10
 
 	// This controls the buffer size for reads to a parquet io.Reader. This value should be small for memory or
 	// disk backed readers, but when the reader is backed by network storage a larger size will be advantageous.

pkg/phlaredb/filter_profiles_bidi.go

@@ -11,21 +11,14 @@ import (
 	"github.com/prometheus/common/model"
 
 	ingestv1 "github.com/grafana/pyroscope/api/gen/proto/go/ingester/v1"
-	typesv1 "github.com/grafana/pyroscope/api/gen/proto/go/types/v1"
 	"github.com/grafana/pyroscope/pkg/iter"
-	phlaremodel "github.com/grafana/pyroscope/pkg/model"
 )
 
 type BidiServerMerge[Res any, Req any] interface {
 	Send(Res) error
 	Receive() (Req, error)
 }
 
-type labelWithIndex struct {
-	phlaremodel.Labels
-	index int
-}
-
 type ProfileWithIndex struct {
 	Profile
 	Index int
@@ -72,8 +65,8 @@ func filterProfiles[B BidiServerMerge[Res, Req], Res filterResponse, Req filterR
 	defer sp.Finish()
 	selection := make([][]Profile, len(profiles))
 	selectProfileResult := &ingestv1.ProfileSets{
-		Profiles:   make([]*ingestv1.SeriesProfile, 0, batchProfileSize),
-		LabelsSets: make([]*typesv1.Labels, 0, batchProfileSize),
+		Profiles:     make([]*ingestv1.SeriesProfile, 0, batchProfileSize),
+		Fingerprints: make([]uint64, 0, batchProfileSize),
 	}
 	its := make([]iter.Iterator[ProfileWithIndex], len(profiles))
 	for i, iter := range profiles {
@@ -92,28 +85,26 @@ func filterProfiles[B BidiServerMerge[Res, Req], Res filterResponse, Req filterR
 			otlog.Int("batch_requested_size", batchProfileSize),
 		)
 
-		seriesByFP := map[model.Fingerprint]labelWithIndex{}
-		selectProfileResult.LabelsSets = selectProfileResult.LabelsSets[:0]
+		seriesByFP := map[model.Fingerprint]int{}
 		selectProfileResult.Profiles = selectProfileResult.Profiles[:0]
+		selectProfileResult.Fingerprints = selectProfileResult.Fingerprints[:0]
 
 		for _, profile := range batch {
 			var ok bool
-			var lblsIdx labelWithIndex
-			lblsIdx, ok = seriesByFP[profile.Fingerprint()]
+			var idx int
+			fp := profile.Fingerprint()
+			idx, ok = seriesByFP[fp]
 			if !ok {
-				lblsIdx = labelWithIndex{
-					Labels: profile.Labels(),
-					index:  len(selectProfileResult.LabelsSets),
-				}
-				seriesByFP[profile.Fingerprint()] = lblsIdx
-				selectProfileResult.LabelsSets = append(selectProfileResult.LabelsSets, &typesv1.Labels{Labels: profile.Labels()})
+				idx = len(selectProfileResult.Fingerprints)
+				seriesByFP[fp] = idx
+				selectProfileResult.Fingerprints = append(selectProfileResult.Fingerprints, uint64(fp))
 			}
 			selectProfileResult.Profiles = append(selectProfileResult.Profiles, &ingestv1.SeriesProfile{
-				LabelIndex: int32(lblsIdx.index),
+				LabelIndex: int32(idx),
 				Timestamp:  int64(profile.Timestamp()),
 			})
-
 		}
+
 		sp.LogFields(otlog.String("msg", "sending batch to client"))
 		var err error
 		switch s := BidiServerMerge[Res, Req](stream).(type) {

pkg/phlaredb/filter_profiles_bidi_test.go

@@ -11,7 +11,6 @@ import (
 	"github.com/stretchr/testify/require"
 
 	ingestv1 "github.com/grafana/pyroscope/api/gen/proto/go/ingester/v1"
-	typesv1 "github.com/grafana/pyroscope/api/gen/proto/go/types/v1"
 	"github.com/grafana/pyroscope/pkg/iter"
 	phlaremodel "github.com/grafana/pyroscope/pkg/model"
 	schemav1 "github.com/grafana/pyroscope/pkg/phlaredb/schemas/v1"
@@ -39,34 +38,36 @@ func TestFilterProfiles(t *testing.T) {
 	require.NoError(t, err)
 	require.Equal(t, 2, len(filtered[0]))
 	require.Equal(t, 3, len(bidi.profilesSent))
-	testhelper.EqualProto(t, []*ingestv1.ProfileSets{
+
+	expectedSent := []*ingestv1.ProfileSets{
 		{
-			LabelsSets: lo.Times(5, func(i int) *typesv1.Labels {
-				return &typesv1.Labels{Labels: phlaremodel.LabelsFromStrings("foo", "bar", "i", fmt.Sprintf("%d", i))}
+			Fingerprints: lo.Times(5, func(i int) uint64 {
+				return phlaremodel.LabelsFromStrings("foo", "bar", "i", fmt.Sprintf("%d", i)).Hash()
 			}),
 			Profiles: lo.Times(5, func(i int) *ingestv1.SeriesProfile {
 				return &ingestv1.SeriesProfile{Timestamp: int64(model.TimeFromUnixNano(int64(i * int(time.Minute)))), LabelIndex: int32(i)}
 			}),
 		},
 		{
-			LabelsSets: lo.Times(5, func(i int) *typesv1.Labels {
-				return &typesv1.Labels{Labels: phlaremodel.LabelsFromStrings("foo", "bar", "i", fmt.Sprintf("%d", i+5))}
+			Fingerprints: lo.Times(5, func(i int) uint64 {
+				return phlaremodel.LabelsFromStrings("foo", "bar", "i", fmt.Sprintf("%d", i+5)).Hash()
 			}),
 			Profiles: lo.Times(5, func(i int) *ingestv1.SeriesProfile {
 				return &ingestv1.SeriesProfile{Timestamp: int64(model.TimeFromUnixNano(int64((i + 5) * int(time.Minute)))), LabelIndex: int32(i)}
 			}),
 		},
 		{
-			LabelsSets: lo.Times(1, func(i int) *typesv1.Labels {
-				return &typesv1.Labels{Labels: phlaremodel.LabelsFromStrings("foo", "bar", "i", fmt.Sprintf("%d", i+10))}
+			Fingerprints: lo.Times(1, func(i int) uint64 {
+				return phlaremodel.LabelsFromStrings("foo", "bar", "i", fmt.Sprintf("%d", i+10)).Hash()
 			}),
 			Profiles: lo.Times(1, func(i int) *ingestv1.SeriesProfile {
 				return &ingestv1.SeriesProfile{Timestamp: int64(model.TimeFromUnixNano(int64((i + 10) * int(time.Minute)))), LabelIndex: int32(i)}
 			}),
 		},
-	}, bidi.profilesSent)
+	}
+	testhelper.EqualProto(t, expectedSent, bidi.profilesSent)
 
-	require.Equal(t, []Profile{
+	expectedFiltered := []Profile{
 		ProfileWithLabels{
 			profile: &schemav1.InMemoryProfile{TimeNanos: int64(5 * int(time.Minute))},
 			lbs:     phlaremodel.LabelsFromStrings("foo", "bar", "i", fmt.Sprintf("%d", 5)),
@@ -77,5 +78,6 @@ func TestFilterProfiles(t *testing.T) {
 			lbs:     phlaremodel.LabelsFromStrings("foo", "bar", "i", fmt.Sprintf("%d", 10)),
 			fp:      model.Fingerprint(phlaremodel.LabelsFromStrings("foo", "bar", "i", fmt.Sprintf("%d", 10)).Hash()),
 		},
-	}, filtered[0])
+	}
+	require.Equal(t, expectedFiltered, filtered[0])
 }

pkg/phlaredb/phlaredb_test.go

@@ -195,7 +195,7 @@ func TestMergeProfilesStacktraces(t *testing.T) {
 		resp, err := bidi.Receive()
 		require.NoError(t, err)
 		require.Nil(t, resp.Result)
-		require.Len(t, resp.SelectedProfiles.LabelsSets, 1)
+		require.Len(t, resp.SelectedProfiles.Fingerprints, 1)
 		require.Len(t, resp.SelectedProfiles.Profiles, 5)
 
 		require.NoError(t, bidi.Send(&ingestv1.MergeProfilesStacktracesRequest{
@@ -325,7 +325,7 @@ func TestMergeProfilesPprof(t *testing.T) {
 		resp, err := bidi.Receive()
 		require.NoError(t, err)
 		require.Nil(t, resp.Result)
-		require.Len(t, resp.SelectedProfiles.LabelsSets, 1)
+		require.Len(t, resp.SelectedProfiles.Fingerprints, 1)
 		require.Len(t, resp.SelectedProfiles.Profiles, 5)
 
 		require.NoError(t, bidi.Send(&ingestv1.MergeProfilesPprofRequest{