feat(v2): metadata string interning (#3744)

api/metastore/v1/types.proto

@@ -2,31 +2,35 @@ syntax = "proto3";
 
 package metastore.v1;
 
-import "types/v1/types.proto";
-
 message BlockList {
   string tenant = 1;
   uint32 shard = 2;
   repeated string blocks = 3;
 }
 
 message BlockMeta {
-  uint64 format_version = 1;
+  uint32 format_version = 1;
+  // Block ID is a unique identifier for the block.
+  // This is the only field that is not included into
+  // the string table.
   string id = 2;
-  int64 min_time = 3;
-  int64 max_time = 4;
-  uint32 shard = 5;
-  uint32 compaction_level = 6;
   // Optional. Empty if compaction level is 0.
-  string tenant_id = 7;
-  repeated Dataset datasets = 8;
+  int32 tenant = 3;
+  uint32 shard = 4;
+  uint32 compaction_level = 5;
+  int64 min_time = 6;
+  int64 max_time = 7;
+  int32 created_by = 8;
   uint64 size = 9;
-  string created_by = 10;
+  repeated Dataset datasets = 10;
+  // String table contains strings of the block.
+  // By convention, the first string is always an empty string.
+  repeated string string_table = 11;
 }
 
 message Dataset {
-  string tenant_id = 1;
-  string name = 2;
+  int32 tenant = 1;
+  int32 name = 2;
   int64 min_time = 3;
   int64 max_time = 4;
 
@@ -44,6 +48,7 @@ message Dataset {
 
   // TODO: Delete. Use labels instead.
   // Profile types present in the tenant service data.
-  repeated string profile_types = 7;
-  repeated types.v1.Labels labels = 8;
+  repeated int32 profile_types = 7;
+  // Length prefixed label key-value pairs.
+  repeated int32 labels = 8;
 }

api/openapiv2/gen/phlare.swagger.json

@@ -602,19 +602,17 @@
       "type": "object",
       "properties": {
         "formatVersion": {
-          "type": "string",
-          "format": "uint64"
+          "type": "integer",
+          "format": "int64"
         },
         "id": {
-          "type": "string"
-        },
-        "minTime": {
           "type": "string",
-          "format": "int64"
+          "description": "Block ID is a unique identifier for the block.\nThis is the only field that is not included into\nthe string table."
         },
-        "maxTime": {
-          "type": "string",
-          "format": "int64"
+        "tenant": {
+          "type": "integer",
+          "format": "int32",
+          "description": "Optional. Empty if compaction level is 0."
         },
         "shard": {
           "type": "integer",
@@ -624,9 +622,21 @@
           "type": "integer",
           "format": "int64"
         },
-        "tenantId": {
+        "minTime": {
           "type": "string",
-          "description": "Optional. Empty if compaction level is 0."
+          "format": "int64"
+        },
+        "maxTime": {
+          "type": "string",
+          "format": "int64"
+        },
+        "createdBy": {
+          "type": "integer",
+          "format": "int32"
+        },
+        "size": {
+          "type": "string",
+          "format": "uint64"
         },
         "datasets": {
           "type": "array",
@@ -635,12 +645,12 @@
             "$ref": "#/definitions/v1Dataset"
           }
         },
-        "size": {
-          "type": "string",
-          "format": "uint64"
-        },
-        "createdBy": {
-          "type": "string"
+        "stringTable": {
+          "type": "array",
+          "items": {
+            "type": "string"
+          },
+          "description": "String table contains strings of the block.\nBy convention, the first string is always an empty string."
         }
       }
     },
@@ -824,11 +834,13 @@
     "v1Dataset": {
       "type": "object",
       "properties": {
-        "tenantId": {
-          "type": "string"
+        "tenant": {
+          "type": "integer",
+          "format": "int32"
         },
         "name": {
-          "type": "string"
+          "type": "integer",
+          "format": "int32"
         },
         "minTime": {
           "type": "string",
@@ -854,16 +866,18 @@
         "profileTypes": {
           "type": "array",
           "items": {
-            "type": "string"
+            "type": "integer",
+            "format": "int32"
           },
           "description": "TODO: Delete. Use labels instead.\nProfile types present in the tenant service data."
         },
         "labels": {
           "type": "array",
           "items": {
-            "type": "object",
-            "$ref": "#/definitions/v1Labels"
-          }
+            "type": "integer",
+            "format": "int32"
+          },
+          "description": "Length prefixed label key-value pairs."
         }
       }
     },

go.mod

@@ -75,7 +75,7 @@ require (
 	github.com/uber/jaeger-client-go v2.30.0+incompatible
 	github.com/valyala/bytebufferpool v1.0.0
 	github.com/xlab/treeprint v1.2.0
-	go.etcd.io/bbolt v1.3.10
+	go.etcd.io/bbolt v1.3.11
 	go.opentelemetry.io/otel v1.30.0
 	go.opentelemetry.io/proto/otlp v1.1.0
 	go.uber.org/atomic v1.11.0

go.sum

@@ -760,8 +760,8 @@ github.com/yuin/goldmark v1.1.27/go.mod h1:3hX8gzYuyVAZsxl0MRgGTJEmQBFcNTphYh9de
 github.com/yuin/goldmark v1.1.32/go.mod h1:3hX8gzYuyVAZsxl0MRgGTJEmQBFcNTphYh9decYSb74=
 github.com/yuin/goldmark v1.2.1/go.mod h1:3hX8gzYuyVAZsxl0MRgGTJEmQBFcNTphYh9decYSb74=
 github.com/yuin/goldmark v1.4.13/go.mod h1:6yULJ656Px+3vBD8DxQVa3kxgyrAnzto9xy5taEt/CY=
-go.etcd.io/bbolt v1.3.10 h1:+BqfJTcCzTItrop8mq/lbzL8wSGtj94UO/3U31shqG0=
-go.etcd.io/bbolt v1.3.10/go.mod h1:bK3UQLPJZly7IlNmV7uVHJDxfe5aK9Ll93e/74Y9oEQ=
+go.etcd.io/bbolt v1.3.11 h1:yGEzV1wPz2yVCLsD8ZAiGHhHVlczyC9d1rP43/VCRJ0=
+go.etcd.io/bbolt v1.3.11/go.mod h1:dksAq7YMXoljX0xu6VF5DMZGbhYYoLUalEiSySYAS4I=
 go.etcd.io/etcd/api/v3 v3.5.7 h1:sbcmosSVesNrWOJ58ZQFitHMdncusIifYcrBfwrlJSY=
 go.etcd.io/etcd/api/v3 v3.5.7/go.mod h1:9qew1gCdDDLu+VwmeG+iFpL+QlpHTo7iubavdVDgCAA=
 go.etcd.io/etcd/client/pkg/v3 v3.5.7 h1:y3kf5Gbp4e4q7egZdn5T7W9TSHUvkClN6u+Rq9mEOmg=

pkg/experiment/query_backend/block/compaction.go → pkg/experiment/block/compaction.go

@@ -2,7 +2,6 @@ package block
 
 import (
 	"context"
-	"crypto/rand"
 	"fmt"
 	"os"
 	"path/filepath"
@@ -12,7 +11,6 @@ import (
 	"sync"
 
 	"github.com/grafana/dskit/multierror"
-	"github.com/oklog/ulid"
 	"github.com/parquet-go/parquet-go"
 	"github.com/prometheus/common/model"
 	"github.com/prometheus/prometheus/storage"
@@ -115,18 +113,17 @@ func PlanCompaction(objects Objects) ([]*CompactionPlan, error) {
 	}
 	level++
 
-	// Assuming that the first block in the job is the oldest one.
-	timestamp := ulid.MustParse(r.meta.Id).Time()
+	g := NewULIDGenerator(objects)
 	m := make(map[string]*CompactionPlan)
 	for _, obj := range objects {
 		for _, s := range obj.meta.Datasets {
-			tm, ok := m[s.TenantId]
+			tm, ok := m[obj.meta.StringTable[s.Tenant]]
 			if !ok {
-				tm = newBlockCompaction(timestamp, s.TenantId, r.meta.Shard, level)
-				m[s.TenantId] = tm
+				tm = newBlockCompaction(g.ULID().String(), obj.meta.StringTable[s.Tenant], r.meta.Shard, level)
+				m[obj.meta.StringTable[s.Tenant]] = tm
 			}
-			sm := tm.addDataset(s)
 			// Bind objects to datasets.
+			sm := tm.addDataset(obj.meta, s)
 			sm.append(NewDataset(s, obj))
 		}
 	}
@@ -135,44 +132,49 @@ func PlanCompaction(objects Objects) ([]*CompactionPlan, error) {
 	for _, tm := range m {
 		ordered = append(ordered, tm)
 		slices.SortFunc(tm.datasets, func(a, b *datasetCompaction) int {
-			return strings.Compare(a.meta.Name, b.meta.Name)
+			return strings.Compare(a.name, b.name)
 		})
 	}
 	slices.SortFunc(ordered, func(a, b *CompactionPlan) int {
-		return strings.Compare(a.tenantID, b.tenantID)
+		return strings.Compare(a.tenant, b.tenant)
 	})
 
 	return ordered, nil
 }
 
 type CompactionPlan struct {
-	tenantID   string
-	datasetMap map[string]*datasetCompaction
+	tenant     string
+	path       string
+	datasetMap map[int32]*datasetCompaction
 	datasets   []*datasetCompaction
 	meta       *metastorev1.BlockMeta
+	strings    *MetadataStrings
 }
 
-func newBlockCompaction(unixMilli uint64, tenantID string, shard uint32, compactionLevel uint32) *CompactionPlan {
-	return &CompactionPlan{
-		tenantID:   tenantID,
-		datasetMap: make(map[string]*datasetCompaction),
-		meta: &metastorev1.BlockMeta{
-			FormatVersion: 1,
-			// TODO(kolesnikovae): Make it deterministic?
-			Id:              ulid.MustNew(unixMilli, rand.Reader).String(),
-			TenantId:        tenantID,
-			Shard:           shard,
-			CompactionLevel: compactionLevel,
-			Datasets:        nil,
-			MinTime:         0,
-			MaxTime:         0,
-			Size:            0,
-		},
-	}
+func newBlockCompaction(
+	id string,
+	tenant string,
+	shard uint32,
+	compactionLevel uint32,
+) *CompactionPlan {
+	p := &CompactionPlan{
+		tenant:     tenant,
+		datasetMap: make(map[int32]*datasetCompaction),
+		strings:    NewMetadataStringTable(),
+	}
+	p.path = BuildObjectPath(tenant, shard, compactionLevel, id)
+	p.meta = &metastorev1.BlockMeta{
+		FormatVersion:   1,
+		Id:              id,
+		Tenant:          p.strings.Put(tenant),
+		Shard:           shard,
+		CompactionLevel: compactionLevel,
+	}
+	return p
 }
 
 func (b *CompactionPlan) Compact(ctx context.Context, dst objstore.Bucket, tmpdir string) (m *metastorev1.BlockMeta, err error) {
-	w := NewBlockWriter(dst, ObjectPath(b.meta), tmpdir)
+	w := NewBlockWriter(dst, b.path, tmpdir)
 	defer func() {
 		err = multierror.New(err, w.Close()).Err()
 	}()
@@ -187,14 +189,17 @@ func (b *CompactionPlan) Compact(ctx context.Context, dst objstore.Bucket, tmpdi
 		return nil, fmt.Errorf("flushing block writer: %w", err)
 	}
 	b.meta.Size = w.Offset()
+	b.meta.StringTable = b.strings.Strings
 	return b.meta, nil
 }
 
-func (b *CompactionPlan) addDataset(s *metastorev1.Dataset) *datasetCompaction {
-	sm, ok := b.datasetMap[s.Name]
+func (b *CompactionPlan) addDataset(md *metastorev1.BlockMeta, s *metastorev1.Dataset) *datasetCompaction {
+	name := b.strings.Put(md.StringTable[s.Name])
+	tenant := b.strings.Put(md.StringTable[s.Tenant])
+	sm, ok := b.datasetMap[name]
 	if !ok {
-		sm = newDatasetCompaction(s.TenantId, s.Name)
-		b.datasetMap[s.Name] = sm
+		sm = b.newDatasetCompaction(tenant, name)
+		b.datasetMap[name] = sm
 		b.datasets = append(b.datasets, sm)
 	}
 	if b.meta.MinTime == 0 || s.MinTime < b.meta.MinTime {
@@ -207,8 +212,12 @@ func (b *CompactionPlan) addDataset(s *metastorev1.Dataset) *datasetCompaction {
 }
 
 type datasetCompaction struct {
+	// Dataset name.
+	name   string
+	parent *CompactionPlan
+
 	meta   *metastorev1.Dataset
-	ptypes map[string]struct{}
+	ptypes map[int32]struct{}
 	path   string // Set at open.
 
 	datasets []*Dataset
@@ -224,12 +233,14 @@ type datasetCompaction struct {
 	flushOnce sync.Once
 }
 
-func newDatasetCompaction(tenantID, name string) *datasetCompaction {
+func (b *CompactionPlan) newDatasetCompaction(tenant, name int32) *datasetCompaction {
 	return &datasetCompaction{
-		ptypes: make(map[string]struct{}, 10),
+		name:   b.strings.Strings[name],
+		parent: b,
+		ptypes: make(map[int32]struct{}, 10),
 		meta: &metastorev1.Dataset{
-			TenantId: tenantID,
-			Name:     name,
+			Tenant: tenant,
+			Name:   name,
 			// Updated at append.
 			MinTime: 0,
 			MaxTime: 0,
@@ -250,7 +261,8 @@ func (m *datasetCompaction) append(s *Dataset) {
 		m.meta.MaxTime = s.meta.MaxTime
 	}
 	for _, pt := range s.meta.ProfileTypes {
-		m.ptypes[pt] = struct{}{}
+		ptn := m.parent.strings.Put(s.obj.meta.StringTable[pt])
+		m.ptypes[ptn] = struct{}{}
 	}
 }
 
@@ -388,11 +400,10 @@ func (m *datasetCompaction) writeTo(w *Writer) (err error) {
 		return err
 	}
 	m.meta.Size = w.Offset() - off
-	m.meta.ProfileTypes = make([]string, 0, len(m.ptypes))
+	m.meta.ProfileTypes = make([]int32, 0, len(m.ptypes))
 	for pt := range m.ptypes {
 		m.meta.ProfileTypes = append(m.meta.ProfileTypes, pt)
 	}
-	sort.Strings(m.meta.ProfileTypes)
 	return nil
 }