diff --git a/chunks.go b/chunks.go
index c0c7954a6e..b3aaeb206d 100644
--- a/chunks.go
+++ b/chunks.go
@@ -36,7 +36,7 @@ type ChunkWriter interface {
 	// must be populated.
 	// After returning successfully, the Ref fields in the ChunkMetas
 	// is set and can be used to retrieve the chunks from the written data.
-	WriteChunks(chunks ...ChunkMeta) error
+	WriteChunks(chunks ...*ChunkMeta) error
 
 	// Close writes any required finalization and closes the resources
 	// associated with the underlying writer.
@@ -156,7 +156,7 @@ func (w *chunkWriter) write(wr io.Writer, b []byte) error {
 	return err
 }
 
-func (w *chunkWriter) WriteChunks(chks ...ChunkMeta) error {
+func (w *chunkWriter) WriteChunks(chks ...*ChunkMeta) error {
 	// Calculate maximum space we need and cut a new segment in case
 	// we don't fit into the current one.
 	maxLen := int64(binary.MaxVarintLen32)
@@ -184,9 +184,7 @@ func (w *chunkWriter) WriteChunks(chks ...ChunkMeta) error {
 	}
 	seq := uint64(w.seq()) << 32
 
-	for i := range chks {
-		chk := &chks[i]
-
+	for _, chk := range chks {
 		chk.Ref = seq | uint64(w.n)
 
 		n = binary.PutUvarint(b, uint64(len(chk.Chunk.Bytes())))
diff --git a/compact.go b/compact.go
index 573e9b440a..a1d7b1e5c5 100644
--- a/compact.go
+++ b/compact.go
@@ -334,7 +334,7 @@ func (c *compactor) populate(blocks []Block, indexw IndexWriter, chunkw ChunkWri
 
 type compactionSet interface {
 	Next() bool
-	At() (labels.Labels, []ChunkMeta)
+	At() (labels.Labels, []*ChunkMeta)
 	Err() error
 }
 
@@ -344,7 +344,7 @@ type compactionSeriesSet struct {
 	chunks ChunkReader
 
 	l   labels.Labels
-	c   []ChunkMeta
+	c   []*ChunkMeta
 	err error
 }
 
@@ -365,9 +365,7 @@ func (c *compactionSeriesSet) Next() bool {
 	if c.err != nil {
 		return false
 	}
-	for i := range c.c {
-		chk := &c.c[i]
-
+	for _, chk := range c.c {
 		chk.Chunk, c.err = c.chunks.Chunk(chk.Ref)
 		if c.err != nil {
 			return false
@@ -384,7 +382,7 @@ func (c *compactionSeriesSet) Err() error {
 	return c.p.Err()
 }
 
-func (c *compactionSeriesSet) At() (labels.Labels, []ChunkMeta) {
+func (c *compactionSeriesSet) At() (labels.Labels, []*ChunkMeta) {
 	return c.l, c.c
 }
 
@@ -393,12 +391,12 @@ type compactionMerger struct {
 
 	aok, bok bool
 	l        labels.Labels
-	c        []ChunkMeta
+	c        []*ChunkMeta
 }
 
 type compactionSeries struct {
 	labels labels.Labels
-	chunks []ChunkMeta
+	chunks []*ChunkMeta
 }
 
 func newCompactionMerger(a, b compactionSet) (*compactionMerger, error) {
@@ -459,7 +457,7 @@ func (c *compactionMerger) Err() error {
 	return c.b.Err()
 }
 
-func (c *compactionMerger) At() (labels.Labels, []ChunkMeta) {
+func (c *compactionMerger) At() (labels.Labels, []*ChunkMeta) {
 	return c.l, c.c
 }
 
diff --git a/head.go b/head.go
index d9b4589ee3..e4c42d8014 100644
--- a/head.go
+++ b/head.go
@@ -480,7 +480,7 @@ func (h *headIndexReader) Postings(name, value string) (Postings, error) {
 }
 
 // Series returns the series for the given reference.
-func (h *headIndexReader) Series(ref uint32) (labels.Labels, []ChunkMeta, error) {
+func (h *headIndexReader) Series(ref uint32) (labels.Labels, []*ChunkMeta, error) {
 	h.mtx.RLock()
 	defer h.mtx.RUnlock()
 
@@ -488,13 +488,13 @@ func (h *headIndexReader) Series(ref uint32) (labels.Labels, []ChunkMeta, error)
 		return nil, nil, ErrNotFound
 	}
 	s := h.series[ref]
-	metas := make([]ChunkMeta, 0, len(s.chunks))
+	metas := make([]*ChunkMeta, 0, len(s.chunks))
 
 	s.mtx.RLock()
 	defer s.mtx.RUnlock()
 
 	for i, c := range s.chunks {
-		metas = append(metas, ChunkMeta{
+		metas = append(metas, &ChunkMeta{
 			MinTime: c.minTime,
 			MaxTime: c.maxTime,
 			Ref:     (uint64(ref) << 32) | uint64(i),
diff --git a/index.go b/index.go
index 23771f035e..f957085763 100644
--- a/index.go
+++ b/index.go
@@ -33,7 +33,7 @@ type IndexWriter interface {
 	// of chunks that the index can reference.
 	// The reference number is used to resolve a series against the postings
 	// list iterator. It only has to be available during the write processing.
-	AddSeries(ref uint32, l labels.Labels, chunks ...ChunkMeta) error
+	AddSeries(ref uint32, l labels.Labels, chunks ...*ChunkMeta) error
 
 	// WriteLabelIndex serializes an index from label names to values.
 	// The passed in values chained tuples of strings of the length of names.
@@ -49,8 +49,8 @@ type IndexWriter interface {
 
 type indexWriterSeries struct {
 	labels labels.Labels
-	chunks []ChunkMeta // series file offset of chunks
-	offset uint32      // index file offset of series reference
+	chunks []*ChunkMeta // series file offset of chunks
+	offset uint32       // index file offset of series reference
 }
 
 // indexWriter implements the IndexWriter interface for the standard
@@ -142,7 +142,7 @@ func (w *indexWriter) writeMeta() error {
 	return w.write(w.bufw, b[:])
 }
 
-func (w *indexWriter) AddSeries(ref uint32, lset labels.Labels, chunks ...ChunkMeta) error {
+func (w *indexWriter) AddSeries(ref uint32, lset labels.Labels, chunks ...*ChunkMeta) error {
 	if _, ok := w.series[ref]; ok {
 		return errors.Errorf("series with reference %d already added", ref)
 	}
@@ -419,7 +419,7 @@ type IndexReader interface {
 	Postings(name, value string) (Postings, error)
 
 	// Series returns the series for the given reference.
-	Series(ref uint32) (labels.Labels, []ChunkMeta, error)
+	Series(ref uint32) (labels.Labels, []*ChunkMeta, error)
 
 	// LabelIndices returns the label pairs for which indices exist.
 	LabelIndices() ([][]string, error)
@@ -599,7 +599,7 @@ func (r *indexReader) LabelIndices() ([][]string, error) {
 	return res, nil
 }
 
-func (r *indexReader) Series(ref uint32) (labels.Labels, []ChunkMeta, error) {
+func (r *indexReader) Series(ref uint32) (labels.Labels, []*ChunkMeta, error) {
 	k, n := binary.Uvarint(r.b[ref:])
 	if n < 1 {
 		return nil, nil, errors.Wrap(errInvalidSize, "number of labels")
@@ -642,7 +642,7 @@ func (r *indexReader) Series(ref uint32) (labels.Labels, []ChunkMeta, error) {
 	}
 
 	b = b[n:]
-	chunks := make([]ChunkMeta, 0, k)
+	chunks := make([]*ChunkMeta, 0, k)
 
 	for i := 0; i < int(k); i++ {
 		firstTime, n := binary.Varint(b)
@@ -663,7 +663,7 @@ func (r *indexReader) Series(ref uint32) (labels.Labels, []ChunkMeta, error) {
 		}
 		b = b[n:]
 
-		chunks = append(chunks, ChunkMeta{
+		chunks = append(chunks, &ChunkMeta{
 			Ref:     o,
 			MinTime: firstTime,
 			MaxTime: lastTime,
diff --git a/index_test.go b/index_test.go
index a9659c95cf..94e62ae556 100644
--- a/index_test.go
+++ b/index_test.go
@@ -121,10 +121,10 @@ func TestPersistence_index_e2e(t *testing.T) {
 
 	// Generate ChunkMetas for every label set.
 	for i, lset := range lbls {
-		var metas []ChunkMeta
+		var metas []*ChunkMeta
 
 		for j := 0; j <= (i % 20); j++ {
-			metas = append(metas, ChunkMeta{
+			metas = append(metas, &ChunkMeta{
 				MinTime: int64(j * 10000),
 				MaxTime: int64((j + 1) * 10000),
 				Ref:     rand.Uint64(),
diff --git a/querier.go b/querier.go
index 31b5d9f1ed..3c42baba32 100644
--- a/querier.go
+++ b/querier.go
@@ -345,7 +345,7 @@ func (s *mergedSeriesSet) Next() bool {
 
 type chunkSeriesSet interface {
 	Next() bool
-	At() (labels.Labels, []ChunkMeta)
+	At() (labels.Labels, []*ChunkMeta)
 	Err() error
 }
 
@@ -357,12 +357,12 @@ type baseChunkSeries struct {
 	absent []string // labels that must be unset in results.
 
 	lset labels.Labels
-	chks []ChunkMeta
+	chks []*ChunkMeta
 	err  error
 }
 
-func (s *baseChunkSeries) At() (labels.Labels, []ChunkMeta) { return s.lset, s.chks }
-func (s *baseChunkSeries) Err() error                       { return s.err }
+func (s *baseChunkSeries) At() (labels.Labels, []*ChunkMeta) { return s.lset, s.chks }
+func (s *baseChunkSeries) Err() error                        { return s.err }
 
 func (s *baseChunkSeries) Next() bool {
 Outer:
@@ -400,20 +400,18 @@ type populatedChunkSeries struct {
 	mint, maxt int64
 
 	err  error
-	chks []ChunkMeta
+	chks []*ChunkMeta
 	lset labels.Labels
 }
 
-func (s *populatedChunkSeries) At() (labels.Labels, []ChunkMeta) { return s.lset, s.chks }
-func (s *populatedChunkSeries) Err() error                       { return s.err }
+func (s *populatedChunkSeries) At() (labels.Labels, []*ChunkMeta) { return s.lset, s.chks }
+func (s *populatedChunkSeries) Err() error                        { return s.err }
 
 func (s *populatedChunkSeries) Next() bool {
 	for s.set.Next() {
 		lset, chks := s.set.At()
 
-		for i := range chks {
-			c := &chks[i]
-
+		for i, c := range chks {
 			if c.MaxTime < s.mint {
 				chks = chks[1:]
 				continue
@@ -468,7 +466,7 @@ func (s *blockSeriesSet) Err() error { return s.err }
 // time series data.
 type chunkSeries struct {
 	labels labels.Labels
-	chunks []ChunkMeta // in-order chunk refs
+	chunks []*ChunkMeta // in-order chunk refs
 }
 
 func (s *chunkSeries) Labels() labels.Labels {
@@ -562,13 +560,13 @@ func (it *chainedSeriesIterator) Err() error {
 // chunkSeriesIterator implements a series iterator on top
 // of a list of time-sorted, non-overlapping chunks.
 type chunkSeriesIterator struct {
-	chunks []ChunkMeta
+	chunks []*ChunkMeta
 
 	i   int
 	cur chunks.Iterator
 }
 
-func newChunkSeriesIterator(cs []ChunkMeta) *chunkSeriesIterator {
+func newChunkSeriesIterator(cs []*ChunkMeta) *chunkSeriesIterator {
 	return &chunkSeriesIterator{
 		chunks: cs,
 		i:      0,