Merge pull request #12304 from bboreham/labels-symboltable

Labels: reduce memory by de-duplicating strings in SymbolTables

.github/workflows/ci.yml

@@ -19,6 +19,7 @@ jobs:
       - run: make GO_ONLY=1 SKIP_GOLANGCI_LINT=1
       - run: go test ./tsdb/ -test.tsdb-isolation=false
       - run: go test --tags=stringlabels ./...
+      - run: go test --tags=dedupelabels ./...
       - run: GOARCH=386 go test ./cmd/prometheus
       - run: make -C documentation/examples/remote_storage
       - run: make -C documentation/examples

cmd/promtool/backfill.go

@@ -127,7 +127,8 @@ func createBlocks(input []byte, mint, maxt, maxBlockDuration int64, maxSamplesIn
 
 			ctx := context.Background()
 			app := w.Appender(ctx)
-			p := textparse.NewOpenMetricsParser(input)
+			symbolTable := labels.NewSymbolTable() // One table per block means it won't grow too large.
+			p := textparse.NewOpenMetricsParser(input, symbolTable)
 			samplesCount := 0
 			for {
 				e, err := p.Next()
@@ -216,7 +217,7 @@ func createBlocks(input []byte, mint, maxt, maxBlockDuration int64, maxSamplesIn
 }
 
 func backfill(maxSamplesInAppender int, input []byte, outputDir string, humanReadable, quiet bool, maxBlockDuration time.Duration) (err error) {
-	p := textparse.NewOpenMetricsParser(input)
+	p := textparse.NewOpenMetricsParser(input, nil) // Don't need a SymbolTable to get max and min timestamps.
 	maxt, mint, err := getMinAndMaxTimestamps(p)
 	if err != nil {
 		return fmt.Errorf("getting min and max timestamp: %w", err)

config/config.go

@@ -82,7 +82,7 @@ func Load(s string, expandExternalLabels bool, logger log.Logger) (*Config, erro
 		return cfg, nil
 	}
 
-	b := labels.ScratchBuilder{}
+	b := labels.NewScratchBuilder(0)
 	cfg.GlobalConfig.ExternalLabels.Range(func(v labels.Label) {
 		newV := os.Expand(v.Value, func(s string) string {
 			if s == "$" {
@@ -97,6 +97,7 @@ func Load(s string, expandExternalLabels bool, logger log.Logger) (*Config, erro
 		if newV != v.Value {
 			level.Debug(logger).Log("msg", "External label replaced", "label", v.Name, "input", v.Value, "output", newV)
 		}
+		// Note newV can be blank. https://github.com/prometheus/prometheus/issues/11024
 		b.Add(v.Name, newV)
 	})
 	cfg.GlobalConfig.ExternalLabels = b.Labels()

model/labels/labels.go

@@ -11,7 +11,7 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 
-//go:build !stringlabels
+//go:build !stringlabels && !dedupelabels
 
 package labels
 
@@ -371,6 +371,25 @@ func (ls Labels) ReleaseStrings(release func(string)) {
 	}
 }
 
+// Builder allows modifying Labels.
+type Builder struct {
+	base Labels
+	del  []string
+	add  []Label
+}
+
+// Reset clears all current state for the builder.
+func (b *Builder) Reset(base Labels) {
+	b.base = base
+	b.del = b.del[:0]
+	b.add = b.add[:0]
+	b.base.Range(func(l Label) {
+		if l.Value == "" {
+			b.del = append(b.del, l.Name)
+		}
+	})
+}
+
 // Labels returns the labels from the builder.
 // If no modifications were made, the original labels are returned.
 func (b *Builder) Labels() Labels {
@@ -401,11 +420,32 @@ type ScratchBuilder struct {
 	add Labels
 }
 
+// Symbol-table is no-op, just for api parity with dedupelabels.
+type SymbolTable struct{}
+
+func NewSymbolTable() *SymbolTable { return nil }
+
+func (t *SymbolTable) Len() int { return 0 }
+
 // NewScratchBuilder creates a ScratchBuilder initialized for Labels with n entries.
 func NewScratchBuilder(n int) ScratchBuilder {
 	return ScratchBuilder{add: make([]Label, 0, n)}
 }
 
+// NewBuilderWithSymbolTable creates a Builder, for api parity with dedupelabels.
+func NewBuilderWithSymbolTable(_ *SymbolTable) *Builder {
+	return NewBuilder(EmptyLabels())
+}
+
+// NewScratchBuilderWithSymbolTable creates a ScratchBuilder, for api parity with dedupelabels.
+func NewScratchBuilderWithSymbolTable(_ *SymbolTable, n int) ScratchBuilder {
+	return NewScratchBuilder(n)
+}
+
+func (b *ScratchBuilder) SetSymbolTable(_ *SymbolTable) {
+	// no-op
+}
+
 func (b *ScratchBuilder) Reset() {
 	b.add = b.add[:0]
 }

model/labels/labels_common.go

@@ -123,13 +123,6 @@ func FromMap(m map[string]string) Labels {
 	return New(l...)
 }
 
-// Builder allows modifying Labels.
-type Builder struct {
-	base Labels
-	del  []string
-	add  []Label
-}
-
 // NewBuilder returns a new LabelsBuilder.
 func NewBuilder(base Labels) *Builder {
 	b := &Builder{
@@ -140,18 +133,6 @@ func NewBuilder(base Labels) *Builder {
 	return b
 }
 
-// Reset clears all current state for the builder.
-func (b *Builder) Reset(base Labels) {
-	b.base = base
-	b.del = b.del[:0]
-	b.add = b.add[:0]
-	b.base.Range(func(l Label) {
-		if l.Value == "" {
-			b.del = append(b.del, l.Name)
-		}
-	})
-}
-
 // Del deletes the label of the given name.
 func (b *Builder) Del(ns ...string) *Builder {
 	for _, n := range ns {

model/labels/labels_dedupelabels.go

@@ -0,0 +1,807 @@
+// Copyright 2024 The Prometheus Authors
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+// http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+//go:build dedupelabels
+
+package labels
+
+import (
+	"bytes"
+	"strings"
+	"sync"
+	"unsafe"
+
+	"github.com/cespare/xxhash/v2"
+	"golang.org/x/exp/slices"
+)
+
+// Labels is implemented by a SymbolTable and string holding name/value
+// pairs encoded as indexes into the table in varint encoding.
+// Names are in alphabetical order.
+type Labels struct {
+	syms *nameTable
+	data string
+}
+
+// Split SymbolTable into the part used by Labels and the part used by Builder.  Only the latter needs the map.
+
+// This part is used by Labels. All fields are immutable after construction.
+type nameTable struct {
+	byNum       []string     // This slice header is never changed, even while we are building the symbol table.
+	symbolTable *SymbolTable // If we need to use it in a Builder.
+}
+
+// SymbolTable is used to map strings into numbers so they can be packed together.
+type SymbolTable struct {
+	mx sync.Mutex
+	*nameTable
+	nextNum int
+	byName  map[string]int
+}
+
+const defaultSymbolTableSize = 1024
+
+func NewSymbolTable() *SymbolTable {
+	t := &SymbolTable{
+		nameTable: &nameTable{byNum: make([]string, defaultSymbolTableSize)},
+		byName:    make(map[string]int, defaultSymbolTableSize),
+	}
+	t.nameTable.symbolTable = t
+	return t
+}
+
+func (t *SymbolTable) Len() int {
+	t.mx.Lock()
+	defer t.mx.Unlock()
+	return len(t.byName)
+}
+
+// ToNum maps a string to an integer, adding the string to the table if it is not already there.
+// Note: copies the string before adding, in case the caller passed part of
+// a buffer that should not be kept alive by this SymbolTable.
+func (t *SymbolTable) ToNum(name string) int {
+	t.mx.Lock()
+	defer t.mx.Unlock()
+	return t.toNumUnlocked(name)
+}
+
+func (t *SymbolTable) toNumUnlocked(name string) int {
+	if i, found := t.byName[name]; found {
+		return i
+	}
+	i := t.nextNum
+	if t.nextNum == cap(t.byNum) {
+		// Name table is full; copy to a new one. Don't touch the existing slice, as nameTable is immutable after construction.
+		newSlice := make([]string, cap(t.byNum)*2)
+		copy(newSlice, t.byNum)
+		t.nameTable = &nameTable{byNum: newSlice, symbolTable: t}
+	}
+	name = strings.Clone(name)
+	t.byNum[i] = name
+	t.byName[name] = i
+	t.nextNum++
+	return i
+}
+
+func (t *SymbolTable) checkNum(name string) (int, bool) {
+	t.mx.Lock()
+	defer t.mx.Unlock()
+	i, bool := t.byName[name]
+	return i, bool
+}
+
+// ToName maps an integer to a string.
+func (t *nameTable) ToName(num int) string {
+	return t.byNum[num]
+}
+
+func decodeVarint(data string, index int) (int, int) {
+	// Fast-path for common case of a single byte, value 0..127.
+	b := data[index]
+	index++
+	if b < 0x80 {
+		return int(b), index
+	}
+	value := int(b & 0x7F)
+	for shift := uint(7); ; shift += 7 {
+		// Just panic if we go of the end of data, since all Labels strings are constructed internally and
+		// malformed data indicates a bug, or memory corruption.
+		b := data[index]
+		index++
+		value |= int(b&0x7F) << shift
+		if b < 0x80 {
+			break
+		}
+	}
+	return value, index
+}
+
+func decodeString(t *nameTable, data string, index int) (string, int) {
+	var num int
+	num, index = decodeVarint(data, index)
+	return t.ToName(num), index
+}
+
+// Bytes returns ls as a byte slice.
+// It uses non-printing characters and so should not be used for printing.
+func (ls Labels) Bytes(buf []byte) []byte {
+	b := bytes.NewBuffer(buf[:0])
+	for i := 0; i < len(ls.data); {
+		if i > 0 {
+			b.WriteByte(seps[0])
+		}
+		var name, value string
+		name, i = decodeString(ls.syms, ls.data, i)
+		value, i = decodeString(ls.syms, ls.data, i)
+		b.WriteString(name)
+		b.WriteByte(seps[0])
+		b.WriteString(value)
+	}
+	return b.Bytes()
+}
+
+// IsZero implements yaml.IsZeroer - if we don't have this then 'omitempty' fields are always omitted.
+func (ls Labels) IsZero() bool {
+	return len(ls.data) == 0
+}
+
+// MatchLabels returns a subset of Labels that matches/does not match with the provided label names based on the 'on' boolean.
+// If on is set to true, it returns the subset of labels that match with the provided label names and its inverse when 'on' is set to false.
+// TODO: This is only used in printing an error message
+func (ls Labels) MatchLabels(on bool, names ...string) Labels {
+	b := NewBuilder(ls)
+	if on {
+		b.Keep(names...)
+	} else {
+		b.Del(MetricName)
+		b.Del(names...)
+	}
+	return b.Labels()
+}
+
+// Hash returns a hash value for the label set.
+// Note: the result is not guaranteed to be consistent across different runs of Prometheus.
+func (ls Labels) Hash() uint64 {
+	// Use xxhash.Sum64(b) for fast path as it's faster.
+	b := make([]byte, 0, 1024)
+	for pos := 0; pos < len(ls.data); {
+		name, newPos := decodeString(ls.syms, ls.data, pos)
+		value, newPos := decodeString(ls.syms, ls.data, newPos)
+		if len(b)+len(name)+len(value)+2 >= cap(b) {
+			// If labels entry is 1KB+, hash the rest of them via Write().
+			h := xxhash.New()
+			_, _ = h.Write(b)
+			for pos < len(ls.data) {
+				name, pos = decodeString(ls.syms, ls.data, pos)
+				value, pos = decodeString(ls.syms, ls.data, pos)
+				_, _ = h.WriteString(name)
+				_, _ = h.Write(seps)
+				_, _ = h.WriteString(value)
+				_, _ = h.Write(seps)
+			}
+			return h.Sum64()
+		}
+
+		b = append(b, name...)
+		b = append(b, seps[0])
+		b = append(b, value...)
+		b = append(b, seps[0])
+		pos = newPos
+	}
+	return xxhash.Sum64(b)
+}
+
+// HashForLabels returns a hash value for the labels matching the provided names.
+// 'names' have to be sorted in ascending order.
+func (ls Labels) HashForLabels(b []byte, names ...string) (uint64, []byte) {
+	b = b[:0]
+	j := 0
+	for i := 0; i < len(ls.data); {
+		var name, value string
+		name, i = decodeString(ls.syms, ls.data, i)
+		value, i = decodeString(ls.syms, ls.data, i)
+		for j < len(names) && names[j] < name {
+			j++
+		}
+		if j == len(names) {
+			break
+		}
+		if name == names[j] {
+			b = append(b, name...)
+			b = append(b, seps[0])
+			b = append(b, value...)
+			b = append(b, seps[0])
+		}
+	}
+
+	return xxhash.Sum64(b), b
+}
+
+// HashWithoutLabels returns a hash value for all labels except those matching
+// the provided names.
+// 'names' have to be sorted in ascending order.
+func (ls Labels) HashWithoutLabels(b []byte, names ...string) (uint64, []byte) {
+	b = b[:0]
+	j := 0
+	for i := 0; i < len(ls.data); {
+		var name, value string
+		name, i = decodeString(ls.syms, ls.data, i)
+		value, i = decodeString(ls.syms, ls.data, i)
+		for j < len(names) && names[j] < name {
+			j++
+		}
+		if name == MetricName || (j < len(names) && name == names[j]) {
+			continue
+		}
+		b = append(b, name...)
+		b = append(b, seps[0])
+		b = append(b, value...)
+		b = append(b, seps[0])
+	}
+	return xxhash.Sum64(b), b
+}
+
+// BytesWithLabels is just as Bytes(), but only for labels matching names.
+// 'names' have to be sorted in ascending order.
+func (ls Labels) BytesWithLabels(buf []byte, names ...string) []byte {
+	b := bytes.NewBuffer(buf[:0])
+	j := 0
+	for pos := 0; pos < len(ls.data); {
+		lName, newPos := decodeString(ls.syms, ls.data, pos)
+		lValue, newPos := decodeString(ls.syms, ls.data, newPos)
+		for j < len(names) && names[j] < lName {
+			j++
+		}
+		if j == len(names) {
+			break
+		}
+		if lName == names[j] {
+			if b.Len() > 1 {
+				b.WriteByte(seps[0])
+			}
+			b.WriteString(lName)
+			b.WriteByte(seps[0])
+			b.WriteString(lValue)
+		}
+		pos = newPos
+	}
+	return b.Bytes()
+}
+
+// BytesWithoutLabels is just as Bytes(), but only for labels not matching names.
+// 'names' have to be sorted in ascending order.
+func (ls Labels) BytesWithoutLabels(buf []byte, names ...string) []byte {
+	b := bytes.NewBuffer(buf[:0])
+	j := 0
+	for pos := 0; pos < len(ls.data); {
+		lName, newPos := decodeString(ls.syms, ls.data, pos)
+		lValue, newPos := decodeString(ls.syms, ls.data, newPos)
+		for j < len(names) && names[j] < lName {
+			j++
+		}
+		if j == len(names) || lName != names[j] {
+			if b.Len() > 1 {
+				b.WriteByte(seps[0])
+			}
+			b.WriteString(lName)
+			b.WriteByte(seps[0])
+			b.WriteString(lValue)
+		}
+		pos = newPos
+	}
+	return b.Bytes()
+}
+
+// Copy returns a copy of the labels.
+func (ls Labels) Copy() Labels {
+	return Labels{syms: ls.syms, data: strings.Clone(ls.data)}
+}
+
+// Get returns the value for the label with the given name.
+// Returns an empty string if the label doesn't exist.
+func (ls Labels) Get(name string) string {
+	if name == "" { // Avoid crash in loop if someone asks for "".
+		return "" // Prometheus does not store blank label names.
+	}
+	for i := 0; i < len(ls.data); {
+		var lName, lValue string
+		lName, i = decodeString(ls.syms, ls.data, i)
+		if lName == name {
+			lValue, _ = decodeString(ls.syms, ls.data, i)
+			return lValue
+		} else if lName[0] > name[0] { // Stop looking if we've gone past.
+			break
+		}
+		_, i = decodeVarint(ls.data, i)
+	}
+	return ""
+}
+
+// Has returns true if the label with the given name is present.
+func (ls Labels) Has(name string) bool {
+	if name == "" { // Avoid crash in loop if someone asks for "".
+		return false // Prometheus does not store blank label names.
+	}
+	for i := 0; i < len(ls.data); {
+		var lName string
+		lName, i = decodeString(ls.syms, ls.data, i)
+		if lName == name {
+			return true
+		} else if lName[0] > name[0] { // Stop looking if we've gone past.
+			break
+		}
+		_, i = decodeVarint(ls.data, i)
+	}
+	return false
+}
+
+// HasDuplicateLabelNames returns whether ls has duplicate label names.
+// It assumes that the labelset is sorted.
+func (ls Labels) HasDuplicateLabelNames() (string, bool) {
+	prevNum := -1
+	for i := 0; i < len(ls.data); {
+		var lNum int
+		lNum, i = decodeVarint(ls.data, i)
+		_, i = decodeVarint(ls.data, i)
+		if lNum == prevNum {
+			return ls.syms.ToName(lNum), true
+		}
+		prevNum = lNum
+	}
+	return "", false
+}
+
+// WithoutEmpty returns the labelset without empty labels.
+// May return the same labelset.
+func (ls Labels) WithoutEmpty() Labels {
+	if ls.IsEmpty() {
+		return ls
+	}
+	// Idea: have a constant symbol for blank, then we don't have to look it up.
+	blank, ok := ls.syms.symbolTable.checkNum("")
+	if !ok { // Symbol table has no entry for blank - none of the values can be blank.
+		return ls
+	}
+	for pos := 0; pos < len(ls.data); {
+		_, newPos := decodeVarint(ls.data, pos)
+		lValue, newPos := decodeVarint(ls.data, newPos)
+		if lValue != blank {
+			pos = newPos
+			continue
+		}
+		// Do not copy the slice until it's necessary.
+		// TODO: could optimise the case where all blanks are at the end.
+		// Note: we size the new buffer on the assumption there is exactly one blank value.
+		buf := make([]byte, pos, pos+(len(ls.data)-newPos))
+		copy(buf, ls.data[:pos]) // copy the initial non-blank labels
+		pos = newPos             // move past the first blank value
+		for pos < len(ls.data) {
+			var newPos int
+			_, newPos = decodeVarint(ls.data, pos)
+			lValue, newPos = decodeVarint(ls.data, newPos)
+			if lValue != blank {
+				buf = append(buf, ls.data[pos:newPos]...)
+			}
+			pos = newPos
+		}
+		return Labels{syms: ls.syms, data: yoloString(buf)}
+	}
+	return ls
+}
+
+// Equal returns whether the two label sets are equal.
+func Equal(a, b Labels) bool {
+	if a.syms == b.syms {
+		return a.data == b.data
+	}
+
+	la, lb := len(a.data), len(b.data)
+	ia, ib := 0, 0
+	for ia < la && ib < lb {
+		var aValue, bValue string
+		aValue, ia = decodeString(a.syms, a.data, ia)
+		bValue, ib = decodeString(b.syms, b.data, ib)
+		if aValue != bValue {
+			return false
+		}
+	}
+	if ia != la || ib != lb {
+		return false
+	}
+	return true
+}
+
+// EmptyLabels returns an empty Labels value, for convenience.
+func EmptyLabels() Labels {
+	return Labels{}
+}
+
+func yoloString(b []byte) string {
+	return *((*string)(unsafe.Pointer(&b)))
+}
+
+// New returns a sorted Labels from the given labels.
+// The caller has to guarantee that all label names are unique.
+// Note this function is not efficient; should not be used in performance-critical places.
+func New(ls ...Label) Labels {
+	slices.SortFunc(ls, func(a, b Label) int { return strings.Compare(a.Name, b.Name) })
+	syms := NewSymbolTable()
+	var stackSpace [16]int
+	size, nums := mapLabelsToNumbers(syms, ls, stackSpace[:])
+	buf := make([]byte, size)
+	marshalNumbersToSizedBuffer(nums, buf)
+	return Labels{syms: syms.nameTable, data: yoloString(buf)}
+}
+
+// FromStrings creates new labels from pairs of strings.
+func FromStrings(ss ...string) Labels {
+	if len(ss)%2 != 0 {
+		panic("invalid number of strings")
+	}
+	ls := make([]Label, 0, len(ss)/2)
+	for i := 0; i < len(ss); i += 2 {
+		ls = append(ls, Label{Name: ss[i], Value: ss[i+1]})
+	}
+
+	return New(ls...)
+}
+
+// Compare compares the two label sets.
+// The result will be 0 if a==b, <0 if a < b, and >0 if a > b.
+func Compare(a, b Labels) int {
+	la, lb := len(a.data), len(b.data)
+	ia, ib := 0, 0
+	for ia < la && ib < lb {
+		var aName, bName string
+		aName, ia = decodeString(a.syms, a.data, ia)
+		bName, ib = decodeString(b.syms, b.data, ib)
+		if aName != bName {
+			if aName < bName {
+				return -1
+			}
+			return 1
+		}
+		var aValue, bValue string
+		aValue, ia = decodeString(a.syms, a.data, ia)
+		bValue, ib = decodeString(b.syms, b.data, ib)
+		if aValue != bValue {
+			if aValue < bValue {
+				return -1
+			}
+			return 1
+		}
+	}
+	// If all labels so far were in common, the set with fewer labels comes first.
+	return (la - ia) - (lb - ib)
+}
+
+// Copy labels from b on top of whatever was in ls previously, reusing memory or expanding if needed.
+func (ls *Labels) CopyFrom(b Labels) {
+	*ls = b // Straightforward memberwise copy is all we need.
+}
+
+// IsEmpty returns true if ls represents an empty set of labels.
+func (ls Labels) IsEmpty() bool {
+	return len(ls.data) == 0
+}
+
+// Len returns the number of labels; it is relatively slow.
+func (ls Labels) Len() int {
+	count := 0
+	for i := 0; i < len(ls.data); {
+		_, i = decodeVarint(ls.data, i)
+		_, i = decodeVarint(ls.data, i)
+		count++
+	}
+	return count
+}
+
+// Range calls f on each label.
+func (ls Labels) Range(f func(l Label)) {
+	for i := 0; i < len(ls.data); {
+		var lName, lValue string
+		lName, i = decodeString(ls.syms, ls.data, i)
+		lValue, i = decodeString(ls.syms, ls.data, i)
+		f(Label{Name: lName, Value: lValue})
+	}
+}
+
+// Validate calls f on each label. If f returns a non-nil error, then it returns that error cancelling the iteration.
+func (ls Labels) Validate(f func(l Label) error) error {
+	for i := 0; i < len(ls.data); {
+		var lName, lValue string
+		lName, i = decodeString(ls.syms, ls.data, i)
+		lValue, i = decodeString(ls.syms, ls.data, i)
+		err := f(Label{Name: lName, Value: lValue})
+		if err != nil {
+			return err
+		}
+	}
+	return nil
+}
+
+// InternStrings calls intern on every string value inside ls, replacing them with what it returns.
+func (ls *Labels) InternStrings(intern func(string) string) {
+	// TODO: remove these calls as there is nothing to do.
+}
+
+// ReleaseStrings calls release on every string value inside ls.
+func (ls Labels) ReleaseStrings(release func(string)) {
+	// TODO: remove these calls as there is nothing to do.
+}
+
+// DropMetricName returns Labels with "__name__" removed.
+func (ls Labels) DropMetricName() Labels {
+	for i := 0; i < len(ls.data); {
+		lName, i2 := decodeString(ls.syms, ls.data, i)
+		_, i2 = decodeVarint(ls.data, i2)
+		if lName == MetricName {
+			if i == 0 { // Make common case fast with no allocations.
+				ls.data = ls.data[i2:]
+			} else {
+				ls.data = ls.data[:i] + ls.data[i2:]
+			}
+			break
+		} else if lName[0] > MetricName[0] { // Stop looking if we've gone past.
+			break
+		}
+		i = i2
+	}
+	return ls
+}
+
+// Builder allows modifying Labels.
+type Builder struct {
+	syms *SymbolTable
+	nums []int
+	base Labels
+	del  []string
+	add  []Label
+}
+
+// NewBuilderWithSymbolTable returns a new LabelsBuilder not based on any labels, but with the SymbolTable.
+func NewBuilderWithSymbolTable(s *SymbolTable) *Builder {
+	return &Builder{
+		syms: s,
+	}
+}
+
+// Reset clears all current state for the builder.
+func (b *Builder) Reset(base Labels) {
+	if base.syms != nil { // If base has a symbol table, use that.
+		b.syms = base.syms.symbolTable
+	} else if b.syms == nil { // Or continue using previous symbol table in builder.
+		b.syms = NewSymbolTable() // Don't do this in performance-sensitive code.
+	}
+
+	b.base = base
+	b.del = b.del[:0]
+	b.add = b.add[:0]
+	base.Range(func(l Label) {
+		if l.Value == "" {
+			b.del = append(b.del, l.Name)
+		}
+	})
+}
+
+// Labels returns the labels from the builder.
+// If no modifications were made, the original labels are returned.
+func (b *Builder) Labels() Labels {
+	if len(b.del) == 0 && len(b.add) == 0 {
+		return b.base
+	}
+
+	slices.SortFunc(b.add, func(a, b Label) int { return strings.Compare(a.Name, b.Name) })
+	slices.Sort(b.del)
+	a, d, newSize := 0, 0, 0
+
+	newSize, b.nums = mapLabelsToNumbers(b.syms, b.add, b.nums)
+	bufSize := len(b.base.data) + newSize
+	buf := make([]byte, 0, bufSize)
+	for pos := 0; pos < len(b.base.data); {
+		oldPos := pos
+		var lName string
+		lName, pos = decodeString(b.base.syms, b.base.data, pos)
+		_, pos = decodeVarint(b.base.data, pos)
+		for d < len(b.del) && b.del[d] < lName {
+			d++
+		}
+		if d < len(b.del) && b.del[d] == lName {
+			continue // This label has been deleted.
+		}
+		for ; a < len(b.add) && b.add[a].Name < lName; a++ {
+			buf = appendLabelTo(b.nums[a*2], b.nums[a*2+1], buf) // Insert label that was not in the base set.
+		}
+		if a < len(b.add) && b.add[a].Name == lName {
+			buf = appendLabelTo(b.nums[a*2], b.nums[a*2+1], buf)
+			a++
+			continue // This label has been replaced.
+		}
+		buf = append(buf, b.base.data[oldPos:pos]...) // If base had a symbol-table we are using it, so we don't need to look up these symbols.
+	}
+	// We have come to the end of the base set; add any remaining labels.
+	for ; a < len(b.add); a++ {
+		buf = appendLabelTo(b.nums[a*2], b.nums[a*2+1], buf)
+	}
+	return Labels{syms: b.syms.nameTable, data: yoloString(buf)}
+}
+
+func marshalNumbersToSizedBuffer(nums []int, data []byte) int {
+	i := len(data)
+	for index := len(nums) - 1; index >= 0; index-- {
+		i = encodeVarint(data, i, nums[index])
+	}
+	return len(data) - i
+}
+
+func sizeVarint(x uint64) (n int) {
+	// Most common case first
+	if x < 1<<7 {
+		return 1
+	}
+	if x >= 1<<56 {
+		return 9
+	}
+	if x >= 1<<28 {
+		x >>= 28
+		n = 4
+	}
+	if x >= 1<<14 {
+		x >>= 14
+		n += 2
+	}
+	if x >= 1<<7 {
+		n++
+	}
+	return n + 1
+}
+
+func encodeVarintSlow(data []byte, offset int, v uint64) int {
+	offset -= sizeVarint(v)
+	base := offset
+	for v >= 1<<7 {
+		data[offset] = uint8(v&0x7f | 0x80)
+		v >>= 7
+		offset++
+	}
+	data[offset] = uint8(v)
+	return base
+}
+
+// Special code for the common case that a value is less than 128
+func encodeVarint(data []byte, offset, v int) int {
+	if v < 1<<7 {
+		offset--
+		data[offset] = uint8(v)
+		return offset
+	}
+	return encodeVarintSlow(data, offset, uint64(v))
+}
+
+// Map all the strings in lbls to the symbol table; return the total size required to hold them and all the individual mappings.
+func mapLabelsToNumbers(t *SymbolTable, lbls []Label, buf []int) (totalSize int, nums []int) {
+	nums = buf[:0]
+	t.mx.Lock()
+	defer t.mx.Unlock()
+	// we just encode name/value/name/value, without any extra tags or length bytes
+	for _, m := range lbls {
+		// strings are encoded as a single varint, the index into the symbol table.
+		i := t.toNumUnlocked(m.Name)
+		nums = append(nums, i)
+		totalSize += sizeVarint(uint64(i))
+		i = t.toNumUnlocked(m.Value)
+		nums = append(nums, i)
+		totalSize += sizeVarint(uint64(i))
+	}
+	return totalSize, nums
+}
+
+func appendLabelTo(nameNum, valueNum int, buf []byte) []byte {
+	size := sizeVarint(uint64(nameNum)) + sizeVarint(uint64(valueNum))
+	sizeRequired := len(buf) + size
+	if cap(buf) >= sizeRequired {
+		buf = buf[:sizeRequired]
+	} else {
+		bufSize := cap(buf)
+		// Double size of buffer each time it needs to grow, to amortise copying cost.
+		for bufSize < sizeRequired {
+			bufSize = bufSize*2 + 1
+		}
+		newBuf := make([]byte, sizeRequired, bufSize)
+		copy(newBuf, buf)
+		buf = newBuf
+	}
+	i := sizeRequired
+	i = encodeVarint(buf, i, valueNum)
+	i = encodeVarint(buf, i, nameNum)
+	return buf
+}
+
+// ScratchBuilder allows efficient construction of a Labels from scratch.
+type ScratchBuilder struct {
+	syms            *SymbolTable
+	nums            []int
+	add             []Label
+	output          Labels
+	overwriteBuffer []byte
+}
+
+// NewScratchBuilder creates a ScratchBuilder initialized for Labels with n entries.
+// Warning: expensive; don't call in tight loops.
+func NewScratchBuilder(n int) ScratchBuilder {
+	return ScratchBuilder{syms: NewSymbolTable(), add: make([]Label, 0, n)}
+}
+
+// NewScratchBuilderWithSymbolTable creates a ScratchBuilder initialized for Labels with n entries.
+func NewScratchBuilderWithSymbolTable(s *SymbolTable, n int) ScratchBuilder {
+	return ScratchBuilder{syms: s, add: make([]Label, 0, n)}
+}
+
+func (b *ScratchBuilder) SetSymbolTable(s *SymbolTable) {
+	b.syms = s
+}
+
+func (b *ScratchBuilder) Reset() {
+	b.add = b.add[:0]
+	b.output = EmptyLabels()
+}
+
+// Add a name/value pair.
+// Note if you Add the same name twice you will get a duplicate label, which is invalid.
+func (b *ScratchBuilder) Add(name, value string) {
+	b.add = append(b.add, Label{Name: name, Value: value})
+}
+
+// Add a name/value pair, using []byte instead of string to reduce memory allocations.
+// The values must remain live until Labels() is called.
+func (b *ScratchBuilder) UnsafeAddBytes(name, value []byte) {
+	b.add = append(b.add, Label{Name: yoloString(name), Value: yoloString(value)})
+}
+
+// Sort the labels added so far by name.
+func (b *ScratchBuilder) Sort() {
+	slices.SortFunc(b.add, func(a, b Label) int { return strings.Compare(a.Name, b.Name) })
+}
+
+// Assign is for when you already have a Labels which you want this ScratchBuilder to return.
+func (b *ScratchBuilder) Assign(l Labels) {
+	b.output = l
+}
+
+// Labels returns the name/value pairs added as a Labels object. Calling Add() after Labels() has no effect.
+// Note: if you want them sorted, call Sort() first.
+func (b *ScratchBuilder) Labels() Labels {
+	if b.output.IsEmpty() {
+		var size int
+		size, b.nums = mapLabelsToNumbers(b.syms, b.add, b.nums)
+		buf := make([]byte, size)
+		marshalNumbersToSizedBuffer(b.nums, buf)
+		b.output = Labels{syms: b.syms.nameTable, data: yoloString(buf)}
+	}
+	return b.output
+}
+
+// Write the newly-built Labels out to ls, reusing an internal buffer.
+// Callers must ensure that there are no other references to ls, or any strings fetched from it.
+func (b *ScratchBuilder) Overwrite(ls *Labels) {
+	var size int
+	size, b.nums = mapLabelsToNumbers(b.syms, b.add, b.nums)
+	if size <= cap(b.overwriteBuffer) {
+		b.overwriteBuffer = b.overwriteBuffer[:size]
+	} else {
+		b.overwriteBuffer = make([]byte, size)
+	}
+	marshalNumbersToSizedBuffer(b.nums, b.overwriteBuffer)
+	ls.syms = b.syms.nameTable
+	ls.data = yoloString(b.overwriteBuffer)
+}

model/labels/labels_stringlabels.go

@@ -188,8 +188,7 @@ func (ls Labels) BytesWithoutLabels(buf []byte, names ...string) []byte {
 
 // Copy returns a copy of the labels.
 func (ls Labels) Copy() Labels {
-	buf := append([]byte{}, ls.data...)
+	return Labels{data: strings.Clone(ls.data)}
-	return Labels{data: yoloString(buf)}
 }
 
 // Get returns the value for the label with the given name.
@@ -458,6 +457,25 @@ func (ls *Labels) InternStrings(intern func(string) string) {
 func (ls Labels) ReleaseStrings(release func(string)) {
 }
 
+// Builder allows modifying Labels.
+type Builder struct {
+	base Labels
+	del  []string
+	add  []Label
+}
+
+// Reset clears all current state for the builder.
+func (b *Builder) Reset(base Labels) {
+	b.base = base
+	b.del = b.del[:0]
+	b.add = b.add[:0]
+	b.base.Range(func(l Label) {
+		if l.Value == "" {
+			b.del = append(b.del, l.Name)
+		}
+	})
+}
+
 // Labels returns the labels from the builder.
 // If no modifications were made, the original labels are returned.
 func (b *Builder) Labels() Labels {
@@ -662,3 +680,24 @@ func (b *ScratchBuilder) Overwrite(ls *Labels) {
 	marshalLabelsToSizedBuffer(b.add, b.overwriteBuffer)
 	ls.data = yoloString(b.overwriteBuffer)
 }
+
+// Symbol-table is no-op, just for api parity with dedupelabels.
+type SymbolTable struct{}
+
+func NewSymbolTable() *SymbolTable { return nil }
+
+func (t *SymbolTable) Len() int { return 0 }
+
+// NewBuilderWithSymbolTable creates a Builder, for api parity with dedupelabels.
+func NewBuilderWithSymbolTable(_ *SymbolTable) *Builder {
+	return NewBuilder(EmptyLabels())
+}
+
+// NewScratchBuilderWithSymbolTable creates a ScratchBuilder, for api parity with dedupelabels.
+func NewScratchBuilderWithSymbolTable(_ *SymbolTable, n int) ScratchBuilder {
+	return NewScratchBuilder(n)
+}
+
+func (b *ScratchBuilder) SetSymbolTable(_ *SymbolTable) {
+	// no-op
+}

model/labels/labels_test.go

@@ -114,7 +114,7 @@ func TestLabels_MatchLabels(t *testing.T) {
 
 	for i, test := range tests {
 		got := labels.MatchLabels(test.on, test.providedNames...)
-		require.Equal(t, test.expected, got, "unexpected labelset for test case %d", i)
+		require.True(t, Equal(test.expected, got), "unexpected labelset for test case %d", i)
 	}
 }
 
@@ -207,7 +207,7 @@ func TestLabels_WithoutEmpty(t *testing.T) {
 		},
 	} {
 		t.Run("", func(t *testing.T) {
-			require.Equal(t, test.expected, test.input.WithoutEmpty())
+			require.True(t, Equal(test.expected, test.input.WithoutEmpty()))
 		})
 	}
 }
@@ -569,6 +569,7 @@ func TestLabels_BytesWithoutLabels(t *testing.T) {
 }
 
 func TestBuilder(t *testing.T) {
+	reuseBuilder := NewBuilderWithSymbolTable(NewSymbolTable())
 	for i, tcase := range []struct {
 		base Labels
 		del  []string
@@ -580,6 +581,11 @@ func TestBuilder(t *testing.T) {
 			base: FromStrings("aaa", "111"),
 			want: FromStrings("aaa", "111"),
 		},
+		{
+			base: EmptyLabels(),
+			set:  []Label{{"aaa", "444"}, {"bbb", "555"}, {"ccc", "666"}},
+			want: FromStrings("aaa", "444", "bbb", "555", "ccc", "666"),
+		},
 		{
 			base: FromStrings("aaa", "111", "bbb", "222", "ccc", "333"),
 			set:  []Label{{"aaa", "444"}, {"bbb", "555"}, {"ccc", "666"}},
@@ -638,8 +644,7 @@ func TestBuilder(t *testing.T) {
 			want: FromStrings("aaa", "111", "ddd", "444"),
 		},
 	} {
-		t.Run(fmt.Sprint(i), func(t *testing.T) {
+		test := func(t *testing.T, b *Builder) {
-			b := NewBuilder(tcase.base)
 			for _, lbl := range tcase.set {
 				b.Set(lbl.Name, lbl.Value)
 			}
@@ -647,7 +652,7 @@ func TestBuilder(t *testing.T) {
 				b.Keep(tcase.keep...)
 			}
 			b.Del(tcase.del...)
-			require.Equal(t, tcase.want, b.Labels())
+			require.True(t, Equal(tcase.want, b.Labels()))
 
 			// Check what happens when we call Range and mutate the builder.
 			b.Range(func(l Label) {
@@ -656,6 +661,18 @@ func TestBuilder(t *testing.T) {
 				}
 			})
 			require.Equal(t, tcase.want.BytesWithoutLabels(nil, "aaa", "bbb"), b.Labels().Bytes(nil))
+		}
+		t.Run(fmt.Sprintf("NewBuilder %d", i), func(t *testing.T) {
+			test(t, NewBuilder(tcase.base))
+		})
+		t.Run(fmt.Sprintf("NewSymbolTable %d", i), func(t *testing.T) {
+			b := NewBuilderWithSymbolTable(NewSymbolTable())
+			b.Reset(tcase.base)
+			test(t, b)
+		})
+		t.Run(fmt.Sprintf("reuseBuilder %d", i), func(t *testing.T) {
+			reuseBuilder.Reset(tcase.base)
+			test(t, reuseBuilder)
 		})
 	}
 	t.Run("set_after_del", func(t *testing.T) {
@@ -694,14 +711,14 @@ func TestScratchBuilder(t *testing.T) {
 		},
 	} {
 		t.Run(fmt.Sprint(i), func(t *testing.T) {
-			b := ScratchBuilder{}
+			b := NewScratchBuilder(len(tcase.add))
 			for _, lbl := range tcase.add {
 				b.Add(lbl.Name, lbl.Value)
 			}
 			b.Sort()
-			require.Equal(t, tcase.want, b.Labels())
+			require.True(t, Equal(tcase.want, b.Labels()))
 			b.Assign(tcase.want)
-			require.Equal(t, tcase.want, b.Labels())
+			require.True(t, Equal(tcase.want, b.Labels()))
 		})
 	}
 }

model/labels/sharding.go

@@ -11,7 +11,7 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 
-//go:build !stringlabels
+//go:build !stringlabels && !dedupelabels
 
 package labels
 

model/labels/sharding_dedupelabels.go

@@ -0,0 +1,52 @@
+// Copyright 2024 The Prometheus Authors
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+// http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+//go:build dedupelabels
+
+package labels
+
+import (
+	"github.com/cespare/xxhash/v2"
+)
+
+// StableHash is a labels hashing implementation which is guaranteed to not change over time.
+// This function should be used whenever labels hashing backward compatibility must be guaranteed.
+func StableHash(ls Labels) uint64 {
+	// Use xxhash.Sum64(b) for fast path as it's faster.
+	b := make([]byte, 0, 1024)
+	for pos := 0; pos < len(ls.data); {
+		name, newPos := decodeString(ls.syms, ls.data, pos)
+		value, newPos := decodeString(ls.syms, ls.data, newPos)
+		if len(b)+len(name)+len(value)+2 >= cap(b) {
+			// If labels entry is 1KB+, hash the rest of them via Write().
+			h := xxhash.New()
+			_, _ = h.Write(b)
+			for pos < len(ls.data) {
+				name, pos = decodeString(ls.syms, ls.data, pos)
+				value, pos = decodeString(ls.syms, ls.data, pos)
+				_, _ = h.WriteString(name)
+				_, _ = h.Write(seps)
+				_, _ = h.WriteString(value)
+				_, _ = h.Write(seps)
+			}
+			return h.Sum64()
+		}
+
+		b = append(b, name...)
+		b = append(b, seps[0])
+		b = append(b, value...)
+		b = append(b, seps[0])
+		pos = newPos
+	}
+	return xxhash.Sum64(b)
+}

model/labels/test_utils.go

@@ -50,7 +50,7 @@ func ReadLabels(fn string, n int) ([]Labels, error) {
 	defer f.Close()
 
 	scanner := bufio.NewScanner(f)
-	b := ScratchBuilder{}
+	b := NewScratchBuilder(0)
 
 	var mets []Labels
 	hashes := map[uint64]struct{}{}

model/textparse/interface.go

@@ -80,22 +80,22 @@ type Parser interface {
 //
 // This function always returns a valid parser, but might additionally
 // return an error if the content type cannot be parsed.
-func New(b []byte, contentType string, parseClassicHistograms bool) (Parser, error) {
+func New(b []byte, contentType string, parseClassicHistograms bool, st *labels.SymbolTable) (Parser, error) {
 	if contentType == "" {
-		return NewPromParser(b), nil
+		return NewPromParser(b, st), nil
 	}
 
 	mediaType, _, err := mime.ParseMediaType(contentType)
 	if err != nil {
-		return NewPromParser(b), err
+		return NewPromParser(b, st), err
 	}
 	switch mediaType {
 	case "application/openmetrics-text":
-		return NewOpenMetricsParser(b), nil
+		return NewOpenMetricsParser(b, st), nil
 	case "application/vnd.google.protobuf":
-		return NewProtobufParser(b, parseClassicHistograms), nil
+		return NewProtobufParser(b, parseClassicHistograms, st), nil
 	default:
-		return NewPromParser(b), nil
+		return NewPromParser(b, st), nil
 	}
 }
 

model/textparse/interface_test.go

@@ -17,6 +17,8 @@ import (
 	"testing"
 
 	"github.com/stretchr/testify/require"
+
+	"github.com/prometheus/prometheus/model/labels"
 )
 
 func TestNewParser(t *testing.T) {
@@ -91,7 +93,7 @@ func TestNewParser(t *testing.T) {
 			tt := tt // Copy to local variable before going parallel.
 			t.Parallel()
 
-			p, err := New([]byte{}, tt.contentType, false)
+			p, err := New([]byte{}, tt.contentType, false, labels.NewSymbolTable())
 			tt.validateParser(t, p)
 			if tt.err == "" {
 				require.NoError(t, err)

model/textparse/openmetricsparse.go

@@ -97,8 +97,11 @@ type OpenMetricsParser struct {
 }
 
 // NewOpenMetricsParser returns a new parser of the byte slice.
-func NewOpenMetricsParser(b []byte) Parser {
+func NewOpenMetricsParser(b []byte, st *labels.SymbolTable) Parser {
-	return &OpenMetricsParser{l: &openMetricsLexer{b: b}}
+	return &OpenMetricsParser{
+		l:       &openMetricsLexer{b: b},
+		builder: labels.NewScratchBuilderWithSymbolTable(st, 16),
+	}
 }
 
 // Series returns the bytes of the series, the timestamp if set, and the value

model/textparse/openmetricsparse_test.go

@@ -247,7 +247,7 @@ foo_total 17.0 1520879607.789 # {id="counter-test"} 5`
 		},
 	}
 
-	p := NewOpenMetricsParser([]byte(input))
+	p := NewOpenMetricsParser([]byte(input), labels.NewSymbolTable())
 	i := 0
 
 	var res labels.Labels
@@ -378,7 +378,7 @@ choices}`, "strange©™\n'quoted' \"name\"", "6"),
 		},
 	}
 
-	p := NewOpenMetricsParser([]byte(input))
+	p := NewOpenMetricsParser([]byte(input), labels.NewSymbolTable())
 	i := 0
 
 	var res labels.Labels
@@ -400,7 +400,7 @@ choices}`, "strange©™\n'quoted' \"name\"", "6"),
 			require.Equal(t, exp[i].m, string(m))
 			require.Equal(t, exp[i].t, ts)
 			require.Equal(t, exp[i].v, v)
-			require.Equal(t, exp[i].lset, res)
+			testutil.RequireEqual(t, exp[i].lset, res)
 			if exp[i].e == nil {
 				require.False(t, found)
 			} else {
@@ -727,7 +727,7 @@ func TestOpenMetricsParseErrors(t *testing.T) {
 	}
 
 	for i, c := range cases {
-		p := NewOpenMetricsParser([]byte(c.input))
+		p := NewOpenMetricsParser([]byte(c.input), labels.NewSymbolTable())
 		var err error
 		for err == nil {
 			_, err = p.Next()
@@ -792,7 +792,7 @@ func TestOMNullByteHandling(t *testing.T) {
 	}
 
 	for i, c := range cases {
-		p := NewOpenMetricsParser([]byte(c.input))
+		p := NewOpenMetricsParser([]byte(c.input), labels.NewSymbolTable())
 		var err error
 		for err == nil {
 			_, err = p.Next()

model/textparse/promparse.go

@@ -166,8 +166,11 @@ type PromParser struct {
 }
 
 // NewPromParser returns a new parser of the byte slice.
-func NewPromParser(b []byte) Parser {
+func NewPromParser(b []byte, st *labels.SymbolTable) Parser {
-	return &PromParser{l: &promlexer{b: append(b, '\n')}}
+	return &PromParser{
+		l:       &promlexer{b: append(b, '\n')},
+		builder: labels.NewScratchBuilderWithSymbolTable(st, 16),
+	}
 }
 
 // Series returns the bytes of the series, the timestamp if set, and the value

model/textparse/promparse_test.go

@@ -178,7 +178,7 @@ testmetric{label="\"bar\""} 1`
 		},
 	}
 
-	p := NewPromParser([]byte(input))
+	p := NewPromParser([]byte(input), labels.NewSymbolTable())
 	i := 0
 
 	var res labels.Labels
@@ -304,7 +304,7 @@ choices}`, "strange©™\n'quoted' \"name\"", "6"),
 		},
 	}
 
-	p := NewPromParser([]byte(input))
+	p := NewPromParser([]byte(input), labels.NewSymbolTable())
 	i := 0
 
 	var res labels.Labels
@@ -325,7 +325,7 @@ choices}`, "strange©™\n'quoted' \"name\"", "6"),
 			require.Equal(t, exp[i].m, string(m))
 			require.Equal(t, exp[i].t, ts)
 			require.Equal(t, exp[i].v, v)
-			require.Equal(t, exp[i].lset, res)
+			testutil.RequireEqual(t, exp[i].lset, res)
 
 		case EntryType:
 			m, typ := p.Type()
@@ -422,7 +422,7 @@ func TestPromParseErrors(t *testing.T) {
 	}
 
 	for i, c := range cases {
-		p := NewPromParser([]byte(c.input))
+		p := NewPromParser([]byte(c.input), labels.NewSymbolTable())
 		var err error
 		for err == nil {
 			_, err = p.Next()
@@ -476,7 +476,7 @@ func TestPromNullByteHandling(t *testing.T) {
 	}
 
 	for i, c := range cases {
-		p := NewPromParser([]byte(c.input))
+		p := NewPromParser([]byte(c.input), labels.NewSymbolTable())
 		var err error
 		for err == nil {
 			_, err = p.Next()
@@ -497,7 +497,7 @@ const (
 )
 
 func BenchmarkParse(b *testing.B) {
-	for parserName, parser := range map[string]func([]byte) Parser{
+	for parserName, parser := range map[string]func([]byte, *labels.SymbolTable) Parser{
 		"prometheus":  NewPromParser,
 		"openmetrics": NewOpenMetricsParser,
 	} {
@@ -516,8 +516,9 @@ func BenchmarkParse(b *testing.B) {
 				b.ReportAllocs()
 				b.ResetTimer()
 
+				st := labels.NewSymbolTable()
 				for i := 0; i < b.N; i += promtestdataSampleCount {
-					p := parser(buf)
+					p := parser(buf, st)
 
 				Outer:
 					for i < b.N {
@@ -544,8 +545,9 @@ func BenchmarkParse(b *testing.B) {
 				b.ReportAllocs()
 				b.ResetTimer()
 
+				st := labels.NewSymbolTable()
 				for i := 0; i < b.N; i += promtestdataSampleCount {
-					p := parser(buf)
+					p := parser(buf, st)
 
 				Outer:
 					for i < b.N {
@@ -577,8 +579,9 @@ func BenchmarkParse(b *testing.B) {
 				b.ReportAllocs()
 				b.ResetTimer()
 
+				st := labels.NewSymbolTable()
 				for i := 0; i < b.N; i += promtestdataSampleCount {
-					p := parser(buf)
+					p := parser(buf, st)
 
 				Outer:
 					for i < b.N {

model/textparse/protobufparse.go

@@ -80,13 +80,14 @@ type ProtobufParser struct {
 }
 
 // NewProtobufParser returns a parser for the payload in the byte slice.
-func NewProtobufParser(b []byte, parseClassicHistograms bool) Parser {
+func NewProtobufParser(b []byte, parseClassicHistograms bool, st *labels.SymbolTable) Parser {
 	return &ProtobufParser{
 		in:                     b,
 		state:                  EntryInvalid,
 		mf:                     &dto.MetricFamily{},
 		metricBytes:            &bytes.Buffer{},
 		parseClassicHistograms: parseClassicHistograms,
+		builder:                labels.NewScratchBuilderWithSymbolTable(st, 16),
 	}
 }
 

model/textparse/protobufparse_test.go

@@ -743,7 +743,7 @@ func TestProtobufParse(t *testing.T) {
 	}{
 		{
 			name:   "ignore classic buckets of native histograms",
-			parser: NewProtobufParser(inputBuf.Bytes(), false),
+			parser: NewProtobufParser(inputBuf.Bytes(), false, labels.NewSymbolTable()),
 			expected: []parseResult{
 				{
 					m:    "go_build_info",
@@ -1280,7 +1280,7 @@ func TestProtobufParse(t *testing.T) {
 		},
 		{
 			name:   "parse classic and native buckets",
-			parser: NewProtobufParser(inputBuf.Bytes(), true),
+			parser: NewProtobufParser(inputBuf.Bytes(), true, labels.NewSymbolTable()),
 			expected: []parseResult{
 				{ // 0
 					m:    "go_build_info",

promql/fuzz.go

@@ -20,6 +20,7 @@ import (
 	"errors"
 	"io"
 
+	"github.com/prometheus/prometheus/model/labels"
 	"github.com/prometheus/prometheus/model/textparse"
 	"github.com/prometheus/prometheus/promql/parser"
 )
@@ -56,8 +57,11 @@ const (
 	maxInputSize = 10240
 )
 
+// Use package-scope symbol table to avoid memory allocation on every fuzzing operation.
+var symbolTable = labels.NewSymbolTable()
+
 func fuzzParseMetricWithContentType(in []byte, contentType string) int {
-	p, warning := textparse.New(in, contentType, false)
+	p, warning := textparse.New(in, contentType, false, symbolTable)
 	if warning != nil {
 		// An invalid content type is being passed, which should not happen
 		// in this context.

rules/alerting.go

@@ -356,6 +356,8 @@ func (r *AlertingRule) Eval(ctx context.Context, ts time.Time, query QueryFunc,
 	// or update the expression value for existing elements.
 	resultFPs := map[uint64]struct{}{}
 
+	lb := labels.NewBuilder(labels.EmptyLabels())
+	sb := labels.NewScratchBuilder(0)
 	var vec promql.Vector
 	alerts := make(map[uint64]*Alert, len(res))
 	for _, smpl := range res {
@@ -391,14 +393,14 @@ func (r *AlertingRule) Eval(ctx context.Context, ts time.Time, query QueryFunc,
 			return result
 		}
 
-		lb := labels.NewBuilder(smpl.Metric).Del(labels.MetricName)
+		lb.Reset(smpl.Metric)
-
+		lb.Del(labels.MetricName)
 		r.labels.Range(func(l labels.Label) {
 			lb.Set(l.Name, expand(l.Value))
 		})
 		lb.Set(labels.AlertName, r.Name())
 
-		sb := labels.ScratchBuilder{}
+		sb.Reset()
 		r.annotations.Range(func(a labels.Label) {
 			sb.Add(a.Name, expand(a.Value))
 		})

scrape/manager_test.go

@@ -528,6 +528,7 @@ scrape_configs:
 		config:      cfg1.ScrapeConfigs[0],
 		client:      http.DefaultClient,
 		metrics:     scrapeManager.metrics,
+		symbolTable: labels.NewSymbolTable(),
 	}
 	scrapeManager.scrapePools = map[string]*scrapePool{
 		"job1": sp,

scrape/scrape.go

@@ -73,6 +73,10 @@ type scrapePool struct {
 	client *http.Client
 	loops  map[uint64]loop
 
+	symbolTable           *labels.SymbolTable
+	lastSymbolTableCheck  time.Time
+	initialSymbolTableLen int
+
 	targetMtx sync.Mutex
 	// activeTargets and loops must always be synchronized to have the same
 	// set of hashes.
@@ -136,6 +140,8 @@ func newScrapePool(cfg *config.ScrapeConfig, app storage.Appendable, offsetSeed
 		client:               client,
 		activeTargets:        map[uint64]*Target{},
 		loops:                map[uint64]loop{},
+		symbolTable:          labels.NewSymbolTable(),
+		lastSymbolTableCheck: time.Now(),
 		logger:               logger,
 		metrics:              metrics,
 		httpOpts:             options.HTTPClientOptions,
@@ -160,6 +166,7 @@ func newScrapePool(cfg *config.ScrapeConfig, app storage.Appendable, offsetSeed
 			func(l labels.Labels) labels.Labels { return mutateReportSampleLabels(l, opts.target) },
 			func(ctx context.Context) storage.Appender { return app.Appender(ctx) },
 			cache,
+			sp.symbolTable,
 			offsetSeed,
 			opts.honorTimestamps,
 			opts.trackTimestampsStaleness,
@@ -348,6 +355,21 @@ func (sp *scrapePool) reload(cfg *config.ScrapeConfig) error {
 	sp.metrics.targetReloadIntervalLength.WithLabelValues(interval.String()).Observe(
 		time.Since(start).Seconds(),
 	)
+
+	// Here we take steps to clear out the symbol table if it has grown a lot.
+	// After waiting some time for things to settle, we take the size of the symbol-table.
+	// If, after some more time, the table has grown to twice that size, we start a new one.
+	const minTimeToCleanSymbolTable = 5 * time.Minute
+	if time.Since(sp.lastSymbolTableCheck) > minTimeToCleanSymbolTable {
+		if sp.initialSymbolTableLen == 0 {
+			sp.initialSymbolTableLen = sp.symbolTable.Len()
+		} else if sp.symbolTable.Len() > 2*sp.initialSymbolTableLen {
+			sp.symbolTable = labels.NewSymbolTable()
+			sp.initialSymbolTableLen = 0
+		}
+		sp.lastSymbolTableCheck = time.Now()
+	}
+
 	return nil
 }
 
@@ -361,7 +383,7 @@ func (sp *scrapePool) Sync(tgs []*targetgroup.Group) {
 	sp.targetMtx.Lock()
 	var all []*Target
 	var targets []*Target
-	lb := labels.NewBuilder(labels.EmptyLabels())
+	lb := labels.NewBuilderWithSymbolTable(sp.symbolTable)
 	sp.droppedTargets = []*Target{}
 	sp.droppedTargetsCount = 0
 	for _, tg := range tgs {
@@ -806,6 +828,7 @@ type scrapeLoop struct {
 	enableCTZeroIngestion    bool
 
 	appender            func(ctx context.Context) storage.Appender
+	symbolTable         *labels.SymbolTable
 	sampleMutator       labelsMutator
 	reportSampleMutator labelsMutator
 
@@ -1085,6 +1108,7 @@ func newScrapeLoop(ctx context.Context,
 	reportSampleMutator labelsMutator,
 	appender func(ctx context.Context) storage.Appender,
 	cache *scrapeCache,
+	symbolTable *labels.SymbolTable,
 	offsetSeed uint64,
 	honorTimestamps bool,
 	trackTimestampsStaleness bool,
@@ -1130,6 +1154,7 @@ func newScrapeLoop(ctx context.Context,
 		buffers:                  buffers,
 		cache:                    cache,
 		appender:                 appender,
+		symbolTable:              symbolTable,
 		sampleMutator:            sampleMutator,
 		reportSampleMutator:      reportSampleMutator,
 		stopped:                  make(chan struct{}),
@@ -1428,7 +1453,7 @@ type appendErrors struct {
 }
 
 func (sl *scrapeLoop) append(app storage.Appender, b []byte, contentType string, ts time.Time) (total, added, seriesAdded int, err error) {
-	p, err := textparse.New(b, contentType, sl.scrapeClassicHistograms)
+	p, err := textparse.New(b, contentType, sl.scrapeClassicHistograms, sl.symbolTable)
 	if err != nil {
 		level.Debug(sl.l).Log(
 			"msg", "Invalid content type on scrape, using prometheus parser as fallback.",
@@ -1778,30 +1803,31 @@ func (sl *scrapeLoop) report(app storage.Appender, start time.Time, duration tim
 	if scrapeErr == nil {
 		health = 1
 	}
+	b := labels.NewBuilderWithSymbolTable(sl.symbolTable)
 
-	if err = sl.addReportSample(app, scrapeHealthMetricName, ts, health); err != nil {
+	if err = sl.addReportSample(app, scrapeHealthMetricName, ts, health, b); err != nil {
 		return
 	}
-	if err = sl.addReportSample(app, scrapeDurationMetricName, ts, duration.Seconds()); err != nil {
+	if err = sl.addReportSample(app, scrapeDurationMetricName, ts, duration.Seconds(), b); err != nil {
 		return
 	}
-	if err = sl.addReportSample(app, scrapeSamplesMetricName, ts, float64(scraped)); err != nil {
+	if err = sl.addReportSample(app, scrapeSamplesMetricName, ts, float64(scraped), b); err != nil {
 		return
 	}
-	if err = sl.addReportSample(app, samplesPostRelabelMetricName, ts, float64(added)); err != nil {
+	if err = sl.addReportSample(app, samplesPostRelabelMetricName, ts, float64(added), b); err != nil {
 		return
 	}
-	if err = sl.addReportSample(app, scrapeSeriesAddedMetricName, ts, float64(seriesAdded)); err != nil {
+	if err = sl.addReportSample(app, scrapeSeriesAddedMetricName, ts, float64(seriesAdded), b); err != nil {
 		return
 	}
 	if sl.reportExtraMetrics {
-		if err = sl.addReportSample(app, scrapeTimeoutMetricName, ts, sl.timeout.Seconds()); err != nil {
+		if err = sl.addReportSample(app, scrapeTimeoutMetricName, ts, sl.timeout.Seconds(), b); err != nil {
 			return
 		}
-		if err = sl.addReportSample(app, scrapeSampleLimitMetricName, ts, float64(sl.sampleLimit)); err != nil {
+		if err = sl.addReportSample(app, scrapeSampleLimitMetricName, ts, float64(sl.sampleLimit), b); err != nil {
 			return
 		}
-		if err = sl.addReportSample(app, scrapeBodySizeBytesMetricName, ts, float64(bytes)); err != nil {
+		if err = sl.addReportSample(app, scrapeBodySizeBytesMetricName, ts, float64(bytes), b); err != nil {
 			return
 		}
 	}
@@ -1812,37 +1838,38 @@ func (sl *scrapeLoop) reportStale(app storage.Appender, start time.Time) (err er
 	ts := timestamp.FromTime(start)
 
 	stale := math.Float64frombits(value.StaleNaN)
+	b := labels.NewBuilder(labels.EmptyLabels())
 
-	if err = sl.addReportSample(app, scrapeHealthMetricName, ts, stale); err != nil {
+	if err = sl.addReportSample(app, scrapeHealthMetricName, ts, stale, b); err != nil {
 		return
 	}
-	if err = sl.addReportSample(app, scrapeDurationMetricName, ts, stale); err != nil {
+	if err = sl.addReportSample(app, scrapeDurationMetricName, ts, stale, b); err != nil {
 		return
 	}
-	if err = sl.addReportSample(app, scrapeSamplesMetricName, ts, stale); err != nil {
+	if err = sl.addReportSample(app, scrapeSamplesMetricName, ts, stale, b); err != nil {
 		return
 	}
-	if err = sl.addReportSample(app, samplesPostRelabelMetricName, ts, stale); err != nil {
+	if err = sl.addReportSample(app, samplesPostRelabelMetricName, ts, stale, b); err != nil {
 		return
 	}
-	if err = sl.addReportSample(app, scrapeSeriesAddedMetricName, ts, stale); err != nil {
+	if err = sl.addReportSample(app, scrapeSeriesAddedMetricName, ts, stale, b); err != nil {
 		return
 	}
 	if sl.reportExtraMetrics {
-		if err = sl.addReportSample(app, scrapeTimeoutMetricName, ts, stale); err != nil {
+		if err = sl.addReportSample(app, scrapeTimeoutMetricName, ts, stale, b); err != nil {
 			return
 		}
-		if err = sl.addReportSample(app, scrapeSampleLimitMetricName, ts, stale); err != nil {
+		if err = sl.addReportSample(app, scrapeSampleLimitMetricName, ts, stale, b); err != nil {
 			return
 		}
-		if err = sl.addReportSample(app, scrapeBodySizeBytesMetricName, ts, stale); err != nil {
+		if err = sl.addReportSample(app, scrapeBodySizeBytesMetricName, ts, stale, b); err != nil {
 			return
 		}
 	}
 	return
 }
 
-func (sl *scrapeLoop) addReportSample(app storage.Appender, s []byte, t int64, v float64) error {
+func (sl *scrapeLoop) addReportSample(app storage.Appender, s []byte, t int64, v float64, b *labels.Builder) error {
 	ce, ok := sl.cache.get(s)
 	var ref storage.SeriesRef
 	var lset labels.Labels
@@ -1853,8 +1880,9 @@ func (sl *scrapeLoop) addReportSample(app storage.Appender, s []byte, t int64, v
 		// The constants are suffixed with the invalid \xff unicode rune to avoid collisions
 		// with scraped metrics in the cache.
 		// We have to drop it when building the actual metric.
-		lset = labels.FromStrings(labels.MetricName, string(s[:len(s)-1]))
+		b.Reset(labels.EmptyLabels())
-		lset = sl.reportSampleMutator(lset)
+		b.Set(labels.MetricName, string(s[:len(s)-1]))
+		lset = sl.reportSampleMutator(b.Labels())
 	}
 
 	ref, err := app.Append(ref, lset, t, v)

scrape/scrape_test.go

@@ -279,6 +279,7 @@ func TestScrapePoolReload(t *testing.T) {
 		logger:        nil,
 		client:        http.DefaultClient,
 		metrics:       newTestScrapeMetrics(t),
+		symbolTable:   labels.NewSymbolTable(),
 	}
 
 	// Reloading a scrape pool with a new scrape configuration must stop all scrape
@@ -361,6 +362,7 @@ func TestScrapePoolReloadPreserveRelabeledIntervalTimeout(t *testing.T) {
 		logger:      nil,
 		client:      http.DefaultClient,
 		metrics:     newTestScrapeMetrics(t),
+		symbolTable: labels.NewSymbolTable(),
 	}
 
 	err := sp.reload(reloadCfg)
@@ -391,6 +393,7 @@ func TestScrapePoolTargetLimit(t *testing.T) {
 		logger:        log.NewNopLogger(),
 		client:        http.DefaultClient,
 		metrics:       newTestScrapeMetrics(t),
+		symbolTable:   labels.NewSymbolTable(),
 	}
 
 	tgs := []*targetgroup.Group{}
@@ -623,6 +626,7 @@ func TestScrapePoolScrapeLoopsStarted(t *testing.T) {
 		logger:        nil,
 		client:        http.DefaultClient,
 		metrics:       newTestScrapeMetrics(t),
+		symbolTable:   labels.NewSymbolTable(),
 	}
 
 	tgs := []*targetgroup.Group{
@@ -660,6 +664,7 @@ func newBasicScrapeLoop(t testing.TB, ctx context.Context, scraper scraper, app
 		nopMutator,
 		app,
 		nil,
+		labels.NewSymbolTable(),
 		0,
 		true,
 		false,
@@ -800,6 +805,7 @@ func TestScrapeLoopRun(t *testing.T) {
 		nopMutator,
 		app,
 		nil,
+		nil,
 		0,
 		true,
 		false,
@@ -942,6 +948,7 @@ func TestScrapeLoopMetadata(t *testing.T) {
 		nopMutator,
 		func(ctx context.Context) storage.Appender { return nopAppender{} },
 		cache,
+		labels.NewSymbolTable(),
 		0,
 		true,
 		false,
@@ -1470,7 +1477,7 @@ func TestScrapeLoopAppendCacheEntryButErrNotFound(t *testing.T) {
 	fakeRef := storage.SeriesRef(1)
 	expValue := float64(1)
 	metric := []byte(`metric{n="1"} 1`)
-	p, warning := textparse.New(metric, "", false)
+	p, warning := textparse.New(metric, "", false, labels.NewSymbolTable())
 	require.NoError(t, warning)
 
 	var lset labels.Labels

scrape/target.go

@@ -169,8 +169,8 @@ func (t *Target) offset(interval time.Duration, offsetSeed uint64) time.Duration
 }
 
 // Labels returns a copy of the set of all public labels of the target.
-func (t *Target) Labels() labels.Labels {
+func (t *Target) Labels(b *labels.ScratchBuilder) labels.Labels {
-	b := labels.NewScratchBuilder(t.labels.Len())
+	b.Reset()
 	t.labels.Range(func(l labels.Label) {
 		if !strings.HasPrefix(l.Name, model.ReservedLabelPrefix) {
 			b.Add(l.Name, l.Value)

scrape/target_test.go

@@ -42,7 +42,8 @@ const (
 func TestTargetLabels(t *testing.T) {
 	target := newTestTarget("example.com:80", 0, labels.FromStrings("job", "some_job", "foo", "bar"))
 	want := labels.FromStrings(model.JobLabel, "some_job", "foo", "bar")
-	got := target.Labels()
+	b := labels.NewScratchBuilder(0)
+	got := target.Labels(&b)
 	require.Equal(t, want, got)
 	i := 0
 	target.LabelsRange(func(l labels.Label) {

storage/remote/codec.go

@@ -176,12 +176,13 @@ func ToQueryResult(ss storage.SeriesSet, sampleLimit int) (*prompb.QueryResult,
 
 // FromQueryResult unpacks and sorts a QueryResult proto.
 func FromQueryResult(sortSeries bool, res *prompb.QueryResult) storage.SeriesSet {
+	b := labels.NewScratchBuilder(0)
 	series := make([]storage.Series, 0, len(res.Timeseries))
 	for _, ts := range res.Timeseries {
 		if err := validateLabelsAndMetricName(ts.Labels); err != nil {
 			return errSeriesSet{err: err}
 		}
-		lbls := labelProtosToLabels(ts.Labels)
+		lbls := labelProtosToLabels(&b, ts.Labels)
 		series = append(series, &concreteSeries{labels: lbls, floats: ts.Samples, histograms: ts.Histograms})
 	}
 
@@ -616,11 +617,11 @@ func FromLabelMatchers(matchers []*prompb.LabelMatcher) ([]*labels.Matcher, erro
 	return result, nil
 }
 
-func exemplarProtoToExemplar(ep prompb.Exemplar) exemplar.Exemplar {
+func exemplarProtoToExemplar(b *labels.ScratchBuilder, ep prompb.Exemplar) exemplar.Exemplar {
 	timestamp := ep.Timestamp
 
 	return exemplar.Exemplar{
-		Labels: labelProtosToLabels(ep.Labels),
+		Labels: labelProtosToLabels(b, ep.Labels),
 		Value:  ep.Value,
 		Ts:     timestamp,
 		HasTs:  timestamp != 0,
@@ -760,8 +761,8 @@ func LabelProtosToMetric(labelPairs []*prompb.Label) model.Metric {
 	return metric
 }
 
-func labelProtosToLabels(labelPairs []prompb.Label) labels.Labels {
+func labelProtosToLabels(b *labels.ScratchBuilder, labelPairs []prompb.Label) labels.Labels {
-	b := labels.ScratchBuilder{}
+	b.Reset()
 	for _, l := range labelPairs {
 		b.Add(l.Name, l.Value)
 	}

storage/remote/codec_test.go

@@ -788,10 +788,11 @@ func (m *mockWriter) Write(p []byte) (n int, err error) {
 type mockChunkSeriesSet struct {
 	chunkedSeries []*prompb.ChunkedSeries
 	index         int
+	builder       labels.ScratchBuilder
 }
 
 func newMockChunkSeriesSet(ss []*prompb.ChunkedSeries) storage.ChunkSeriesSet {
-	return &mockChunkSeriesSet{chunkedSeries: ss, index: -1}
+	return &mockChunkSeriesSet{chunkedSeries: ss, index: -1, builder: labels.NewScratchBuilder(0)}
 }
 
 func (c *mockChunkSeriesSet) Next() bool {
@@ -801,7 +802,7 @@ func (c *mockChunkSeriesSet) Next() bool {
 
 func (c *mockChunkSeriesSet) At() storage.ChunkSeries {
 	return &storage.ChunkSeriesEntry{
-		Lset: labelProtosToLabels(c.chunkedSeries[c.index].Labels),
+		Lset: labelProtosToLabels(&c.builder, c.chunkedSeries[c.index].Labels),
 		ChunkIteratorFn: func(chunks.Iterator) chunks.Iterator {
 			return &mockChunkIterator{
 				chunks: c.chunkedSeries[c.index].Chunks,

storage/remote/queue_manager_test.go

@@ -163,16 +163,23 @@ func TestSampleDelivery(t *testing.T) {
 	}
 }
 
-func TestMetadataDelivery(t *testing.T) {
+func newTestClientAndQueueManager(t testing.TB, flushDeadline time.Duration) (*TestWriteClient, *QueueManager) {
 	c := NewTestWriteClient()
-
-	dir := t.TempDir()
-
 	cfg := config.DefaultQueueConfig
 	mcfg := config.DefaultMetadataConfig
+	return c, newTestQueueManager(t, cfg, mcfg, flushDeadline, c)
+}
 
+func newTestQueueManager(t testing.TB, cfg config.QueueConfig, mcfg config.MetadataConfig, deadline time.Duration, c WriteClient) *QueueManager {
+	dir := t.TempDir()
 	metrics := newQueueManagerMetrics(nil, "", "")
-	m := NewQueueManager(metrics, nil, nil, nil, dir, newEWMARate(ewmaWeight, shardUpdateDuration), cfg, mcfg, labels.EmptyLabels(), nil, c, defaultFlushDeadline, newPool(), newHighestTimestampMetric(), nil, false, false)
+	m := NewQueueManager(metrics, nil, nil, nil, dir, newEWMARate(ewmaWeight, shardUpdateDuration), cfg, mcfg, labels.EmptyLabels(), nil, c, deadline, newPool(), newHighestTimestampMetric(), nil, false, false)
+
+	return m
+}
+
+func TestMetadataDelivery(t *testing.T) {
+	c, m := newTestClientAndQueueManager(t, defaultFlushDeadline)
 	m.Start()
 	defer m.Stop()
 
@@ -192,7 +199,7 @@ func TestMetadataDelivery(t *testing.T) {
 	require.Len(t, c.receivedMetadata, numMetadata)
 	// One more write than the rounded qoutient should be performed in order to get samples that didn't
 	// fit into MaxSamplesPerSend.
-	require.Equal(t, numMetadata/mcfg.MaxSamplesPerSend+1, c.writesReceived)
+	require.Equal(t, numMetadata/config.DefaultMetadataConfig.MaxSamplesPerSend+1, c.writesReceived)
 	// Make sure the last samples were sent.
 	require.Equal(t, c.receivedMetadata[metadata[len(metadata)-1].Metric][0].MetricFamilyName, metadata[len(metadata)-1].Metric)
 }
@@ -201,17 +208,13 @@ func TestSampleDeliveryTimeout(t *testing.T) {
 	// Let's send one less sample than batch size, and wait the timeout duration
 	n := 9
 	samples, series := createTimeseries(n, n)
-	c := NewTestWriteClient()
 
 	cfg := config.DefaultQueueConfig
-	mcfg := config.DefaultMetadataConfig
 	cfg.MaxShards = 1
 	cfg.BatchSendDeadline = model.Duration(100 * time.Millisecond)
 
-	dir := t.TempDir()
+	c := NewTestWriteClient()
-
+	m := newTestQueueManager(t, cfg, config.DefaultMetadataConfig, defaultFlushDeadline, c)
-	metrics := newQueueManagerMetrics(nil, "", "")
-	m := NewQueueManager(metrics, nil, nil, nil, dir, newEWMARate(ewmaWeight, shardUpdateDuration), cfg, mcfg, labels.EmptyLabels(), nil, c, defaultFlushDeadline, newPool(), newHighestTimestampMetric(), nil, false, false)
 	m.StoreSeries(series, 0)
 	m.Start()
 	defer m.Stop()
@@ -244,16 +247,8 @@ func TestSampleDeliveryOrder(t *testing.T) {
 		})
 	}
 
-	c := NewTestWriteClient()
+	c, m := newTestClientAndQueueManager(t, defaultFlushDeadline)
 	c.expectSamples(samples, series)
-
-	dir := t.TempDir()
-
-	cfg := config.DefaultQueueConfig
-	mcfg := config.DefaultMetadataConfig
-
-	metrics := newQueueManagerMetrics(nil, "", "")
-	m := NewQueueManager(metrics, nil, nil, nil, dir, newEWMARate(ewmaWeight, shardUpdateDuration), cfg, mcfg, labels.EmptyLabels(), nil, c, defaultFlushDeadline, newPool(), newHighestTimestampMetric(), nil, false, false)
 	m.StoreSeries(series, 0)
 
 	m.Start()
@@ -267,13 +262,10 @@ func TestShutdown(t *testing.T) {
 	deadline := 1 * time.Second
 	c := NewTestBlockedWriteClient()
 
-	dir := t.TempDir()
-
 	cfg := config.DefaultQueueConfig
 	mcfg := config.DefaultMetadataConfig
-	metrics := newQueueManagerMetrics(nil, "", "")
 
-	m := NewQueueManager(metrics, nil, nil, nil, dir, newEWMARate(ewmaWeight, shardUpdateDuration), cfg, mcfg, labels.EmptyLabels(), nil, c, deadline, newPool(), newHighestTimestampMetric(), nil, false, false)
+	m := newTestQueueManager(t, cfg, mcfg, deadline, c)
 	n := 2 * config.DefaultQueueConfig.MaxSamplesPerSend
 	samples, series := createTimeseries(n, n)
 	m.StoreSeries(series, 0)
@@ -306,12 +298,10 @@ func TestSeriesReset(t *testing.T) {
 	numSegments := 4
 	numSeries := 25
 
-	dir := t.TempDir()
-
 	cfg := config.DefaultQueueConfig
 	mcfg := config.DefaultMetadataConfig
-	metrics := newQueueManagerMetrics(nil, "", "")
+	m := newTestQueueManager(t, cfg, mcfg, deadline, c)
-	m := NewQueueManager(metrics, nil, nil, nil, dir, newEWMARate(ewmaWeight, shardUpdateDuration), cfg, mcfg, labels.EmptyLabels(), nil, c, deadline, newPool(), newHighestTimestampMetric(), nil, false, false)
+
 	for i := 0; i < numSegments; i++ {
 		series := []record.RefSeries{}
 		for j := 0; j < numSeries; j++ {
@@ -330,17 +320,12 @@ func TestReshard(t *testing.T) {
 	nSamples := config.DefaultQueueConfig.Capacity * size
 	samples, series := createTimeseries(nSamples, nSeries)
 
-	c := NewTestWriteClient()
-	c.expectSamples(samples, series)
-
 	cfg := config.DefaultQueueConfig
-	mcfg := config.DefaultMetadataConfig
 	cfg.MaxShards = 1
 
-	dir := t.TempDir()
+	c := NewTestWriteClient()
-
+	m := newTestQueueManager(t, cfg, config.DefaultMetadataConfig, defaultFlushDeadline, c)
-	metrics := newQueueManagerMetrics(nil, "", "")
+	c.expectSamples(samples, series)
-	m := NewQueueManager(metrics, nil, nil, nil, dir, newEWMARate(ewmaWeight, shardUpdateDuration), cfg, mcfg, labels.EmptyLabels(), nil, c, defaultFlushDeadline, newPool(), newHighestTimestampMetric(), nil, false, false)
 	m.StoreSeries(series, 0)
 
 	m.Start()
@@ -363,7 +348,7 @@ func TestReshard(t *testing.T) {
 	c.waitForExpectedData(t)
 }
 
-func TestReshardRaceWithStop(*testing.T) {
+func TestReshardRaceWithStop(t *testing.T) {
 	c := NewTestWriteClient()
 	var m *QueueManager
 	h := sync.Mutex{}
@@ -375,8 +360,7 @@ func TestReshardRaceWithStop(*testing.T) {
 	exitCh := make(chan struct{})
 	go func() {
 		for {
-			metrics := newQueueManagerMetrics(nil, "", "")
+			m = newTestQueueManager(t, cfg, mcfg, defaultFlushDeadline, c)
-			m = NewQueueManager(metrics, nil, nil, nil, "", newEWMARate(ewmaWeight, shardUpdateDuration), cfg, mcfg, labels.EmptyLabels(), nil, c, defaultFlushDeadline, newPool(), newHighestTimestampMetric(), nil, false, false)
 			m.Start()
 			h.Unlock()
 			h.Lock()
@@ -410,8 +394,7 @@ func TestReshardPartialBatch(t *testing.T) {
 	flushDeadline := 10 * time.Millisecond
 	cfg.BatchSendDeadline = model.Duration(batchSendDeadline)
 
-	metrics := newQueueManagerMetrics(nil, "", "")
+	m := newTestQueueManager(t, cfg, mcfg, flushDeadline, c)
-	m := NewQueueManager(metrics, nil, nil, nil, t.TempDir(), newEWMARate(ewmaWeight, shardUpdateDuration), cfg, mcfg, labels.EmptyLabels(), nil, c, flushDeadline, newPool(), newHighestTimestampMetric(), nil, false, false)
 	m.StoreSeries(series, 0)
 
 	m.Start()
@@ -454,9 +437,7 @@ func TestQueueFilledDeadlock(t *testing.T) {
 	batchSendDeadline := time.Millisecond
 	cfg.BatchSendDeadline = model.Duration(batchSendDeadline)
 
-	metrics := newQueueManagerMetrics(nil, "", "")
+	m := newTestQueueManager(t, cfg, mcfg, flushDeadline, c)
-
-	m := NewQueueManager(metrics, nil, nil, nil, t.TempDir(), newEWMARate(ewmaWeight, shardUpdateDuration), cfg, mcfg, labels.EmptyLabels(), nil, c, flushDeadline, newPool(), newHighestTimestampMetric(), nil, false, false)
 	m.StoreSeries(series, 0)
 	m.Start()
 	defer m.Stop()
@@ -479,11 +460,7 @@ func TestQueueFilledDeadlock(t *testing.T) {
 }
 
 func TestReleaseNoninternedString(t *testing.T) {
-	cfg := config.DefaultQueueConfig
+	_, m := newTestClientAndQueueManager(t, defaultFlushDeadline)
-	mcfg := config.DefaultMetadataConfig
-	metrics := newQueueManagerMetrics(nil, "", "")
-	c := NewTestWriteClient()
-	m := NewQueueManager(metrics, nil, nil, nil, "", newEWMARate(ewmaWeight, shardUpdateDuration), cfg, mcfg, labels.EmptyLabels(), nil, c, defaultFlushDeadline, newPool(), newHighestTimestampMetric(), nil, false, false)
 	m.Start()
 	defer m.Stop()
 
@@ -525,12 +502,8 @@ func TestShouldReshard(t *testing.T) {
 		},
 	}
 
-	cfg := config.DefaultQueueConfig
-	mcfg := config.DefaultMetadataConfig
 	for _, c := range cases {
-		metrics := newQueueManagerMetrics(nil, "", "")
+		_, m := newTestClientAndQueueManager(t, defaultFlushDeadline)
-		client := NewTestWriteClient()
-		m := NewQueueManager(metrics, nil, nil, nil, "", newEWMARate(ewmaWeight, shardUpdateDuration), cfg, mcfg, labels.EmptyLabels(), nil, client, defaultFlushDeadline, newPool(), newHighestTimestampMetric(), nil, false, false)
 		m.numShards = c.startingShards
 		m.dataIn.incr(c.samplesIn)
 		m.dataOut.incr(c.samplesOut)
@@ -550,7 +523,7 @@ func TestShouldReshard(t *testing.T) {
 func createTimeseries(numSamples, numSeries int, extraLabels ...labels.Label) ([]record.RefSample, []record.RefSeries) {
 	samples := make([]record.RefSample, 0, numSamples)
 	series := make([]record.RefSeries, 0, numSeries)
-	lb := labels.ScratchBuilder{}
+	lb := labels.NewScratchBuilder(1 + len(extraLabels))
 	for i := 0; i < numSeries; i++ {
 		name := fmt.Sprintf("test_metric_%d", i)
 		for j := 0; j < numSamples; j++ {
@@ -787,9 +760,10 @@ func (c *TestWriteClient) Store(_ context.Context, req []byte, _ int) error {
 	if err := proto.Unmarshal(reqBuf, &reqProto); err != nil {
 		return err
 	}
+	builder := labels.NewScratchBuilder(0)
 	count := 0
 	for _, ts := range reqProto.Timeseries {
-		labels := labelProtosToLabels(ts.Labels)
+		labels := labelProtosToLabels(&builder, ts.Labels)
 		seriesName := labels.Get("__name__")
 		for _, sample := range ts.Samples {
 			count++
@@ -904,10 +878,7 @@ func BenchmarkSampleSend(b *testing.B) {
 	cfg.MinShards = 20
 	cfg.MaxShards = 20
 
-	dir := b.TempDir()
+	m := newTestQueueManager(b, cfg, mcfg, defaultFlushDeadline, c)
-
-	metrics := newQueueManagerMetrics(nil, "", "")
-	m := NewQueueManager(metrics, nil, nil, nil, dir, newEWMARate(ewmaWeight, shardUpdateDuration), cfg, mcfg, labels.EmptyLabels(), nil, c, defaultFlushDeadline, newPool(), newHighestTimestampMetric(), nil, false, false)
 	m.StoreSeries(series, 0)
 
 	// These should be received by the client.
@@ -1083,15 +1054,9 @@ func TestProcessExternalLabels(t *testing.T) {
 }
 
 func TestCalculateDesiredShards(t *testing.T) {
-	c := NewTestWriteClient()
 	cfg := config.DefaultQueueConfig
-	mcfg := config.DefaultMetadataConfig
+	_, m := newTestClientAndQueueManager(t, defaultFlushDeadline)
-
+	samplesIn := m.dataIn
-	dir := t.TempDir()
-
-	metrics := newQueueManagerMetrics(nil, "", "")
-	samplesIn := newEWMARate(ewmaWeight, shardUpdateDuration)
-	m := NewQueueManager(metrics, nil, nil, nil, dir, samplesIn, cfg, mcfg, labels.EmptyLabels(), nil, c, defaultFlushDeadline, newPool(), newHighestTimestampMetric(), nil, false, false)
 
 	// Need to start the queue manager so the proper metrics are initialized.
 	// However we can stop it right away since we don't need to do any actual
@@ -1160,15 +1125,8 @@ func TestCalculateDesiredShards(t *testing.T) {
 }
 
 func TestCalculateDesiredShardsDetail(t *testing.T) {
-	c := NewTestWriteClient()
+	_, m := newTestClientAndQueueManager(t, defaultFlushDeadline)
-	cfg := config.DefaultQueueConfig
+	samplesIn := m.dataIn
-	mcfg := config.DefaultMetadataConfig
-
-	dir := t.TempDir()
-
-	metrics := newQueueManagerMetrics(nil, "", "")
-	samplesIn := newEWMARate(ewmaWeight, shardUpdateDuration)
-	m := NewQueueManager(metrics, nil, nil, nil, dir, samplesIn, cfg, mcfg, labels.EmptyLabels(), nil, c, defaultFlushDeadline, newPool(), newHighestTimestampMetric(), nil, false, false)
 
 	for _, tc := range []struct {
 		name            string
@@ -1393,10 +1351,7 @@ func TestDropOldTimeSeries(t *testing.T) {
 	mcfg := config.DefaultMetadataConfig
 	cfg.MaxShards = 1
 	cfg.SampleAgeLimit = model.Duration(60 * time.Second)
-	dir := t.TempDir()
+	m := newTestQueueManager(t, cfg, mcfg, defaultFlushDeadline, c)
-
-	metrics := newQueueManagerMetrics(nil, "", "")
-	m := NewQueueManager(metrics, nil, nil, nil, dir, newEWMARate(ewmaWeight, shardUpdateDuration), cfg, mcfg, labels.EmptyLabels(), nil, c, defaultFlushDeadline, newPool(), newHighestTimestampMetric(), nil, false, false)
 	m.StoreSeries(series, 0)
 
 	m.Start()
@@ -1416,7 +1371,7 @@ func createTimeseriesWithOldSamples(numSamples, numSeries int, extraLabels ...la
 	newSamples := make([]record.RefSample, 0, numSamples)
 	samples := make([]record.RefSample, 0, numSamples)
 	series := make([]record.RefSeries, 0, numSeries)
-	lb := labels.ScratchBuilder{}
+	lb := labels.NewScratchBuilder(1 + len(extraLabels))
 	for i := 0; i < numSeries; i++ {
 		name := fmt.Sprintf("test_metric_%d", i)
 		// We create half of the samples in the past.

storage/remote/read_test.go

@@ -195,6 +195,7 @@ func TestSeriesSetFilter(t *testing.T) {
 type mockedRemoteClient struct {
 	got   *prompb.Query
 	store []*prompb.TimeSeries
+	b     labels.ScratchBuilder
 }
 
 func (c *mockedRemoteClient) Read(_ context.Context, query *prompb.Query) (*prompb.QueryResult, error) {
@@ -210,7 +211,7 @@ func (c *mockedRemoteClient) Read(_ context.Context, query *prompb.Query) (*prom
 
 	q := &prompb.QueryResult{}
 	for _, s := range c.store {
-		l := labelProtosToLabels(s.Labels)
+		l := labelProtosToLabels(&c.b, s.Labels)
 		var notMatch bool
 
 		for _, m := range matchers {
@@ -242,6 +243,7 @@ func TestSampleAndChunkQueryableClient(t *testing.T) {
 			{Labels: []prompb.Label{{Name: "a", Value: "b3"}, {Name: "region", Value: "us"}}},
 			{Labels: []prompb.Label{{Name: "a", Value: "b2"}, {Name: "region", Value: "europe"}}},
 		},
+		b: labels.NewScratchBuilder(0),
 	}
 
 	for _, tc := range []struct {

storage/remote/write_handler.go

@@ -25,6 +25,7 @@ import (
 	"github.com/prometheus/client_golang/prometheus"
 
 	"github.com/prometheus/prometheus/model/exemplar"
+	"github.com/prometheus/prometheus/model/labels"
 	"github.com/prometheus/prometheus/prompb"
 	"github.com/prometheus/prometheus/storage"
 	otlptranslator "github.com/prometheus/prometheus/storage/remote/otlptranslator/prometheusremotewrite"
@@ -112,9 +113,10 @@ func (h *writeHandler) write(ctx context.Context, req *prompb.WriteRequest) (err
 		err = app.Commit()
 	}()
 
+	b := labels.NewScratchBuilder(0)
 	var exemplarErr error
 	for _, ts := range req.Timeseries {
-		labels := labelProtosToLabels(ts.Labels)
+		labels := labelProtosToLabels(&b, ts.Labels)
 		if !labels.IsValid() {
 			level.Warn(h.logger).Log("msg", "Invalid metric names or labels", "got", labels.String())
 			samplesWithInvalidLabels++
@@ -137,7 +139,7 @@ func (h *writeHandler) write(ctx context.Context, req *prompb.WriteRequest) (err
 		}
 
 		for _, ep := range ts.Exemplars {
-			e := exemplarProtoToExemplar(ep)
+			e := exemplarProtoToExemplar(&b, ep)
 
 			_, exemplarErr = app.AppendExemplar(0, labels, e)
 			exemplarErr = h.checkAppendExemplarError(exemplarErr, e, &outOfOrderExemplarErrs)

storage/remote/write_handler_test.go

@@ -55,18 +55,19 @@ func TestRemoteWriteHandler(t *testing.T) {
 	resp := recorder.Result()
 	require.Equal(t, http.StatusNoContent, resp.StatusCode)
 
+	b := labels.NewScratchBuilder(0)
 	i := 0
 	j := 0
 	k := 0
 	for _, ts := range writeRequestFixture.Timeseries {
-		labels := labelProtosToLabels(ts.Labels)
+		labels := labelProtosToLabels(&b, ts.Labels)
 		for _, s := range ts.Samples {
 			requireEqual(t, mockSample{labels, s.Timestamp, s.Value}, appendable.samples[i])
 			i++
 		}
 
 		for _, e := range ts.Exemplars {
-			exemplarLabels := labelProtosToLabels(e.Labels)
+			exemplarLabels := labelProtosToLabels(&b, e.Labels)
 			requireEqual(t, mockExemplar{labels, exemplarLabels, e.Timestamp, e.Value}, appendable.exemplars[j])
 			j++
 		}

tsdb/agent/db.go

@@ -417,7 +417,8 @@ func (db *DB) replayWAL() error {
 
 func (db *DB) loadWAL(r *wlog.Reader, multiRef map[chunks.HeadSeriesRef]chunks.HeadSeriesRef) (err error) {
 	var (
-		dec     record.Decoder
+		syms    = labels.NewSymbolTable() // One table for the whole WAL.
+		dec     = record.NewDecoder(syms)
 		lastRef = chunks.HeadSeriesRef(db.nextRef.Load())
 
 		decoded = make(chan interface{}, 10)

tsdb/agent/db_test.go

@@ -184,7 +184,7 @@ func TestCommit(t *testing.T) {
 	// Read records from WAL and check for expected count of series, samples, and exemplars.
 	var (
 		r   = wlog.NewReader(sr)
-		dec record.Decoder
+		dec = record.NewDecoder(labels.NewSymbolTable())
 
 		walSeriesCount, walSamplesCount, walExemplarsCount, walHistogramCount, walFloatHistogramCount int
 	)
@@ -293,7 +293,7 @@ func TestRollback(t *testing.T) {
 	// Read records from WAL and check for expected count of series and samples.
 	var (
 		r   = wlog.NewReader(sr)
-		dec record.Decoder
+		dec = record.NewDecoder(labels.NewSymbolTable())
 
 		walSeriesCount, walSamplesCount, walHistogramCount, walFloatHistogramCount, walExemplarsCount int
 	)
@@ -737,7 +737,7 @@ func TestStorage_DuplicateExemplarsIgnored(t *testing.T) {
 	defer sr.Close()
 	r := wlog.NewReader(sr)
 
-	var dec record.Decoder
+	dec := record.NewDecoder(labels.NewSymbolTable())
 	for r.Next() {
 		rec := r.Record()
 		if dec.Type(rec) == record.Exemplars {

tsdb/db_test.go

@@ -4031,7 +4031,7 @@ func TestOOOWALWrite(t *testing.T) {
 
 		var (
 			records []interface{}
-			dec     record.Decoder
+			dec     record.Decoder = record.NewDecoder(labels.NewSymbolTable())
 		)
 		for r.Next() {
 			rec := r.Record()
@@ -5390,7 +5390,7 @@ func TestWBLAndMmapReplay(t *testing.T) {
 		require.NoError(t, err)
 		sr, err := wlog.NewSegmentsReader(originalWblDir)
 		require.NoError(t, err)
-		var dec record.Decoder
+		dec := record.NewDecoder(labels.NewSymbolTable())
 		r, markers, addedRecs := wlog.NewReader(sr), 0, 0
 		for r.Next() {
 			rec := r.Record()

tsdb/head.go

@@ -717,6 +717,7 @@ func (h *Head) Init(minValidTime int64) error {
 
 	h.startWALReplayStatus(startFrom, endAt)
 
+	syms := labels.NewSymbolTable() // One table for the whole WAL.
 	multiRef := map[chunks.HeadSeriesRef]chunks.HeadSeriesRef{}
 	if err == nil && startFrom >= snapIdx {
 		sr, err := wlog.NewSegmentsReader(dir)
@@ -731,7 +732,7 @@ func (h *Head) Init(minValidTime int64) error {
 
 		// A corrupted checkpoint is a hard error for now and requires user
 		// intervention. There's likely little data that can be recovered anyway.
-		if err := h.loadWAL(wlog.NewReader(sr), multiRef, mmappedChunks, oooMmappedChunks); err != nil {
+		if err := h.loadWAL(wlog.NewReader(sr), syms, multiRef, mmappedChunks, oooMmappedChunks); err != nil {
 			return fmt.Errorf("backfill checkpoint: %w", err)
 		}
 		h.updateWALReplayStatusRead(startFrom)
@@ -764,7 +765,7 @@ func (h *Head) Init(minValidTime int64) error {
 		if err != nil {
 			return fmt.Errorf("segment reader (offset=%d): %w", offset, err)
 		}
-		err = h.loadWAL(wlog.NewReader(sr), multiRef, mmappedChunks, oooMmappedChunks)
+		err = h.loadWAL(wlog.NewReader(sr), syms, multiRef, mmappedChunks, oooMmappedChunks)
 		if err := sr.Close(); err != nil {
 			level.Warn(h.logger).Log("msg", "Error while closing the wal segments reader", "err", err)
 		}
@@ -792,7 +793,7 @@ func (h *Head) Init(minValidTime int64) error {
 			}
 
 			sr := wlog.NewSegmentBufReader(s)
-			err = h.loadWBL(wlog.NewReader(sr), multiRef, lastMmapRef)
+			err = h.loadWBL(wlog.NewReader(sr), syms, multiRef, lastMmapRef)
 			if err := sr.Close(); err != nil {
 				level.Warn(h.logger).Log("msg", "Error while closing the wbl segments reader", "err", err)
 			}

tsdb/head_test.go

@@ -172,7 +172,7 @@ func readTestWAL(t testing.TB, dir string) (recs []interface{}) {
 		require.NoError(t, sr.Close())
 	}()
 
-	var dec record.Decoder
+	dec := record.NewDecoder(labels.NewSymbolTable())
 	r := wlog.NewReader(sr)
 
 	for r.Next() {

tsdb/head_wal.go

@@ -52,7 +52,7 @@ type histogramRecord struct {
 	fh  *histogram.FloatHistogram
 }
 
-func (h *Head) loadWAL(r *wlog.Reader, multiRef map[chunks.HeadSeriesRef]chunks.HeadSeriesRef, mmappedChunks, oooMmappedChunks map[chunks.HeadSeriesRef][]*mmappedChunk) (err error) {
+func (h *Head) loadWAL(r *wlog.Reader, syms *labels.SymbolTable, multiRef map[chunks.HeadSeriesRef]chunks.HeadSeriesRef, mmappedChunks, oooMmappedChunks map[chunks.HeadSeriesRef][]*mmappedChunk) (err error) {
 	// Track number of samples that referenced a series we don't know about
 	// for error reporting.
 	var unknownRefs atomic.Uint64
@@ -69,7 +69,6 @@ func (h *Head) loadWAL(r *wlog.Reader, multiRef map[chunks.HeadSeriesRef]chunks.
 		processors     = make([]walSubsetProcessor, concurrency)
 		exemplarsInput chan record.RefExemplar
 
-		dec             record.Decoder
 		shards          = make([][]record.RefSample, concurrency)
 		histogramShards = make([][]histogramRecord, concurrency)
 
@@ -137,6 +136,7 @@ func (h *Head) loadWAL(r *wlog.Reader, multiRef map[chunks.HeadSeriesRef]chunks.
 	go func() {
 		defer close(decoded)
 		var err error
+		dec := record.NewDecoder(syms)
 		for r.Next() {
 			rec := r.Record()
 			switch dec.Type(rec) {
@@ -645,7 +645,7 @@ func (wp *walSubsetProcessor) processWALSamples(h *Head, mmappedChunks, oooMmapp
 	return unknownRefs, unknownHistogramRefs, mmapOverlappingChunks
 }
 
-func (h *Head) loadWBL(r *wlog.Reader, multiRef map[chunks.HeadSeriesRef]chunks.HeadSeriesRef, lastMmapRef chunks.ChunkDiskMapperRef) (err error) {
+func (h *Head) loadWBL(r *wlog.Reader, syms *labels.SymbolTable, multiRef map[chunks.HeadSeriesRef]chunks.HeadSeriesRef, lastMmapRef chunks.ChunkDiskMapperRef) (err error) {
 	// Track number of samples, m-map markers, that referenced a series we don't know about
 	// for error reporting.
 	var unknownRefs, mmapMarkerUnknownRefs atomic.Uint64
@@ -657,7 +657,7 @@ func (h *Head) loadWBL(r *wlog.Reader, multiRef map[chunks.HeadSeriesRef]chunks.
 		concurrency = h.opts.WALReplayConcurrency
 		processors  = make([]wblSubsetProcessor, concurrency)
 
-		dec    record.Decoder
+		dec    = record.NewDecoder(syms)
 		shards = make([][]record.RefSample, concurrency)
 
 		decodedCh   = make(chan interface{}, 10)
@@ -1360,7 +1360,8 @@ func (h *Head) loadChunkSnapshot() (int, int, map[chunks.HeadSeriesRef]*memSerie
 		errChan          = make(chan error, concurrency)
 		refSeries        map[chunks.HeadSeriesRef]*memSeries
 		exemplarBuf      []record.RefExemplar
-		dec              record.Decoder
+		syms             = labels.NewSymbolTable() // New table for the whole snapshot.
+		dec              = record.NewDecoder(syms)
 	)
 
 	wg.Add(concurrency)

tsdb/index/index.go

@@ -1118,6 +1118,7 @@ type Reader struct {
 	symbols     *Symbols
 	nameSymbols map[uint32]string // Cache of the label name symbol lookups,
 	// as there are not many and they are half of all lookups.
+	st *labels.SymbolTable // TODO: see if we can merge this with nameSymbols.
 
 	dec *Decoder
 
@@ -1177,6 +1178,7 @@ func newReader(b ByteSlice, c io.Closer) (*Reader, error) {
 		b:        b,
 		c:        c,
 		postings: map[string][]postingOffset{},
+		st:       labels.NewSymbolTable(),
 	}
 
 	// Verify header.
@@ -1653,6 +1655,8 @@ func (r *Reader) Series(id storage.SeriesRef, builder *labels.ScratchBuilder, ch
 	if d.Err() != nil {
 		return d.Err()
 	}
+	builder.SetSymbolTable(r.st)
+	builder.Reset()
 	err := r.dec.Series(d.Get(), builder, chks)
 	if err != nil {
 		return fmt.Errorf("read series: %w", err)

tsdb/record/record.go

@@ -192,11 +192,14 @@ type RefMmapMarker struct {
 }
 
 // Decoder decodes series, sample, metadata and tombstone records.
-// The zero value is ready to use.
 type Decoder struct {
 	builder labels.ScratchBuilder
 }
 
+func NewDecoder(t *labels.SymbolTable) Decoder { // FIXME remove t
+	return Decoder{builder: labels.NewScratchBuilder(0)}
+}
+
 // Type returns the type of the record.
 // Returns RecordUnknown if no valid record type is found.
 func (d *Decoder) Type(rec []byte) Type {

tsdb/record/record_test.go

@@ -29,7 +29,7 @@ import (
 
 func TestRecord_EncodeDecode(t *testing.T) {
 	var enc Encoder
-	var dec Decoder
+	dec := NewDecoder(labels.NewSymbolTable())
 
 	series := []RefSeries{
 		{
@@ -187,7 +187,7 @@ func TestRecord_EncodeDecode(t *testing.T) {
 // Bugfix check for pull/521 and pull/523.
 func TestRecord_Corrupted(t *testing.T) {
 	var enc Encoder
-	var dec Decoder
+	dec := NewDecoder(labels.NewSymbolTable())
 
 	t.Run("Test corrupted series record", func(t *testing.T) {
 		series := []RefSeries{

tsdb/wal.go

@@ -31,6 +31,7 @@ import (
 	"github.com/go-kit/log/level"
 	"github.com/prometheus/client_golang/prometheus"
 
+	"github.com/prometheus/prometheus/model/labels"
 	"github.com/prometheus/prometheus/storage"
 	"github.com/prometheus/prometheus/tsdb/chunks"
 	"github.com/prometheus/prometheus/tsdb/encoding"
@@ -859,6 +860,7 @@ func newWALReader(files []*segmentFile, l log.Logger) *walReader {
 		files:  files,
 		buf:    make([]byte, 0, 128*4096),
 		crc32:  newCRC32(),
+		dec:    record.NewDecoder(labels.NewSymbolTable()),
 	}
 }
 

tsdb/wal_test.go

@@ -510,7 +510,7 @@ func TestMigrateWAL_Fuzz(t *testing.T) {
 
 	r := wlog.NewReader(sr)
 	var res []interface{}
-	var dec record.Decoder
+	dec := record.NewDecoder(labels.NewSymbolTable())
 
 	for r.Next() {
 		rec := r.Record()

tsdb/wlog/checkpoint.go

@@ -28,6 +28,7 @@ import (
 	"github.com/go-kit/log/level"
 	"golang.org/x/exp/slices"
 
+	"github.com/prometheus/prometheus/model/labels"
 	"github.com/prometheus/prometheus/tsdb/chunks"
 	tsdb_errors "github.com/prometheus/prometheus/tsdb/errors"
 	"github.com/prometheus/prometheus/tsdb/fileutil"
@@ -154,7 +155,8 @@ func Checkpoint(logger log.Logger, w *WL, from, to int, keep func(id chunks.Head
 		tstones          []tombstones.Stone
 		exemplars        []record.RefExemplar
 		metadata         []record.RefMetadata
-		dec              record.Decoder
+		st               = labels.NewSymbolTable() // Needed for decoding; labels do not outlive this function.
+		dec              = record.NewDecoder(st)
 		enc              record.Encoder
 		buf              []byte
 		recs             [][]byte

tsdb/wlog/checkpoint_test.go

@@ -237,7 +237,7 @@ func TestCheckpoint(t *testing.T) {
 			require.NoError(t, err)
 			defer sr.Close()
 
-			var dec record.Decoder
+			dec := record.NewDecoder(labels.NewSymbolTable())
 			var series []record.RefSeries
 			var metadata []record.RefMetadata
 			r := NewReader(sr)

tsdb/wlog/watcher.go

@@ -29,6 +29,7 @@ import (
 	"github.com/prometheus/client_golang/prometheus"
 	"golang.org/x/exp/slices"
 
+	"github.com/prometheus/prometheus/model/labels"
 	"github.com/prometheus/prometheus/model/timestamp"
 	"github.com/prometheus/prometheus/tsdb/record"
 )
@@ -532,7 +533,7 @@ func (w *Watcher) garbageCollectSeries(segmentNum int) error {
 // Also used with readCheckpoint - implements segmentReadFn.
 func (w *Watcher) readSegment(r *LiveReader, segmentNum int, tail bool) error {
 	var (
-		dec                   record.Decoder
+		dec                   = record.NewDecoder(labels.NewSymbolTable()) // One table per WAL segment means it won't grow indefinitely.
 		series                []record.RefSeries
 		samples               []record.RefSample
 		samplesToSend         []record.RefSample
@@ -669,7 +670,7 @@ func (w *Watcher) readSegment(r *LiveReader, segmentNum int, tail bool) error {
 // Used with readCheckpoint - implements segmentReadFn.
 func (w *Watcher) readSegmentForGC(r *LiveReader, segmentNum int, _ bool) error {
 	var (
-		dec    record.Decoder
+		dec    = record.NewDecoder(labels.NewSymbolTable()) // Needed for decoding; labels do not outlive this function.
 		series []record.RefSeries
 	)
 	for r.Next() && !isClosed(w.quit) {

web/api/v1/api.go

@@ -1009,6 +1009,7 @@ func (api *API) targets(r *http.Request) apiFuncResult {
 		targetsActive := api.targetRetriever(r.Context()).TargetsActive()
 		activeKeys, numTargets := sortKeys(targetsActive)
 		res.ActiveTargets = make([]*Target, 0, numTargets)
+		builder := labels.NewScratchBuilder(0)
 
 		for _, key := range activeKeys {
 			if scrapePool != "" && key != scrapePool {
@@ -1025,7 +1026,7 @@ func (api *API) targets(r *http.Request) apiFuncResult {
 
 				res.ActiveTargets = append(res.ActiveTargets, &Target{
 					DiscoveredLabels: target.DiscoveredLabels(),
-					Labels:           target.Labels(),
+					Labels:           target.Labels(&builder),
 					ScrapePool:       key,
 					ScrapeURL:        target.URL().String(),
 					GlobalURL:        globalURL.String(),
@@ -1101,6 +1102,7 @@ func (api *API) targetMetadata(r *http.Request) apiFuncResult {
 		}
 	}
 
+	builder := labels.NewScratchBuilder(0)
 	metric := r.FormValue("metric")
 	res := []metricMetadata{}
 	for _, tt := range api.targetRetriever(r.Context()).TargetsActive() {
@@ -1108,15 +1110,16 @@ func (api *API) targetMetadata(r *http.Request) apiFuncResult {
 			if limit >= 0 && len(res) >= limit {
 				break
 			}
+			targetLabels := t.Labels(&builder)
 			// Filter targets that don't satisfy the label matchers.
-			if matchTarget != "" && !matchLabels(t.Labels(), matchers) {
+			if matchTarget != "" && !matchLabels(targetLabels, matchers) {
 				continue
 			}
 			// If no metric is specified, get the full list for the target.
 			if metric == "" {
 				for _, md := range t.ListMetadata() {
 					res = append(res, metricMetadata{
-						Target: t.Labels(),
+						Target: targetLabels,
 						Metric: md.Metric,
 						Type:   md.Type,
 						Help:   md.Help,
@@ -1128,7 +1131,7 @@ func (api *API) targetMetadata(r *http.Request) apiFuncResult {
 			// Get metadata for the specified metric.
 			if md, ok := t.GetMetadata(metric); ok {
 				res = append(res, metricMetadata{
-					Target: t.Labels(),
+					Target: targetLabels,
 					Type:   md.Type,
 					Help:   md.Help,
 					Unit:   md.Unit,

web/federate_test.go

@@ -391,7 +391,7 @@ func TestFederationWithNativeHistograms(t *testing.T) {
 	body, err := io.ReadAll(res.Body)
 	require.NoError(t, err)
 
-	p := textparse.NewProtobufParser(body, false)
+	p := textparse.NewProtobufParser(body, false, labels.NewSymbolTable())
 	var actVec promql.Vector
 	metricFamilies := 0
 	l := labels.Labels{}