diff --git a/CHANGELOG.md b/CHANGELOG.md
index 7c40d5e..1964566 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -1,5 +1,9 @@
 # Changelog
 
+## Unreleased
+
+- Cluster and Ring pipelines process commands for each node in its own goroutine.
+
 ## 6.14
 
 - Added Options.MinIdleConns.
diff --git a/cluster.go b/cluster.go
index 8cbd447..658f6c9 100644
--- a/cluster.go
+++ b/cluster.go
@@ -1241,7 +1241,8 @@ func (c *ClusterClient) WrapProcessPipeline(
 }
 
 func (c *ClusterClient) defaultProcessPipeline(cmds []Cmder) error {
-	cmdsMap, err := c.mapCmdsByNode(cmds)
+	cmdsMap := newCmdsMap()
+	err := c.mapCmdsByNode(cmds, cmdsMap)
 	if err != nil {
 		setCmdsErr(cmds, err)
 		return err
@@ -1252,51 +1253,35 @@ func (c *ClusterClient) defaultProcessPipeline(cmds []Cmder) error {
 			time.Sleep(c.retryBackoff(attempt))
 		}
 
-		failedCmds := make(map[*clusterNode][]Cmder)
-
+		failedCmds := newCmdsMap()
 		var wg sync.WaitGroup
-		var lock sync.RWMutex
-		for node, cmds := range cmdsMap {
+
+		for node, cmds := range cmdsMap.m {
 			wg.Add(1)
 			go func(node *clusterNode, cmds []Cmder) {
 				defer wg.Done()
 
-				failedCmdsTmp := make(map[*clusterNode][]Cmder)
-
 				cn, err := node.Client.getConn()
 				if err != nil {
 					if err == pool.ErrClosed {
-						c.remapCmds(cmds, failedCmdsTmp)
+						c.mapCmdsByNode(cmds, failedCmds)
 					} else {
 						setCmdsErr(cmds, err)
 					}
+					return
+				}
 
+				err = c.pipelineProcessCmds(node, cn, cmds, failedCmds)
+				if err == nil || internal.IsRedisError(err) {
+					node.Client.connPool.Put(cn)
 				} else {
-					err = c.pipelineProcessCmds(node, cn, cmds, failedCmdsTmp)
-					if err == nil || internal.IsRedisError(err) {
-						node.Client.connPool.Put(cn)
-					} else {
-						node.Client.connPool.Remove(cn)
-					}
+					node.Client.connPool.Remove(cn)
 				}
-
-				if len(failedCmdsTmp) > 0 {
-					for node, cs := range failedCmdsTmp {
-						lock.Lock()
-						if _, ok := failedCmds[node]; ok {
-							failedCmds[node] = append(failedCmds[node], cs...)
-						} else {
-							failedCmds[node] = cs
-						}
-						lock.Unlock()
-					}
-				}
-
 			}(node, cmds)
 		}
-		wg.Wait()
 
-		if len(failedCmds) == 0 {
+		wg.Wait()
+		if len(failedCmds.m) == 0 {
 			break
 		}
 		cmdsMap = failedCmds
@@ -1305,14 +1290,24 @@ func (c *ClusterClient) defaultProcessPipeline(cmds []Cmder) error {
 	return cmdsFirstErr(cmds)
 }
 
-func (c *ClusterClient) mapCmdsByNode(cmds []Cmder) (map[*clusterNode][]Cmder, error) {
+type cmdsMap struct {
+	mu sync.Mutex
+	m  map[*clusterNode][]Cmder
+}
+
+func newCmdsMap() *cmdsMap {
+	return &cmdsMap{
+		m: make(map[*clusterNode][]Cmder),
+	}
+}
+
+func (c *ClusterClient) mapCmdsByNode(cmds []Cmder, cmdsMap *cmdsMap) error {
 	state, err := c.state.Get()
 	if err != nil {
 		setCmdsErr(cmds, err)
-		return nil, err
+		return err
 	}
 
-	cmdsMap := make(map[*clusterNode][]Cmder)
 	cmdsAreReadOnly := c.cmdsAreReadOnly(cmds)
 	for _, cmd := range cmds {
 		var node *clusterNode
@@ -1324,11 +1319,13 @@ func (c *ClusterClient) mapCmdsByNode(cmds []Cmder) (map[*clusterNode][]Cmder, e
 			node, err = state.slotMasterNode(slot)
 		}
 		if err != nil {
-			return nil, err
+			return err
 		}
-		cmdsMap[node] = append(cmdsMap[node], cmd)
+		cmdsMap.mu.Lock()
+		cmdsMap.m[node] = append(cmdsMap.m[node], cmd)
+		cmdsMap.mu.Unlock()
 	}
-	return cmdsMap, nil
+	return nil
 }
 
 func (c *ClusterClient) cmdsAreReadOnly(cmds []Cmder) bool {
@@ -1341,27 +1338,17 @@ func (c *ClusterClient) cmdsAreReadOnly(cmds []Cmder) bool {
 	return true
 }
 
-func (c *ClusterClient) remapCmds(cmds []Cmder, failedCmds map[*clusterNode][]Cmder) {
-	remappedCmds, err := c.mapCmdsByNode(cmds)
-	if err != nil {
-		setCmdsErr(cmds, err)
-		return
-	}
-
-	for node, cmds := range remappedCmds {
-		failedCmds[node] = cmds
-	}
-}
-
 func (c *ClusterClient) pipelineProcessCmds(
-	node *clusterNode, cn *pool.Conn, cmds []Cmder, failedCmds map[*clusterNode][]Cmder,
+	node *clusterNode, cn *pool.Conn, cmds []Cmder, failedCmds *cmdsMap,
 ) error {
 	err := cn.WithWriter(c.opt.WriteTimeout, func(wr *proto.Writer) error {
 		return writeCmd(wr, cmds...)
 	})
 	if err != nil {
 		setCmdsErr(cmds, err)
-		failedCmds[node] = cmds
+		failedCmds.mu.Lock()
+		failedCmds.m[node] = cmds
+		failedCmds.mu.Unlock()
 		return err
 	}
 
@@ -1372,7 +1359,7 @@ func (c *ClusterClient) pipelineProcessCmds(
 }
 
 func (c *ClusterClient) pipelineReadCmds(
-	rd *proto.Reader, cmds []Cmder, failedCmds map[*clusterNode][]Cmder,
+	rd *proto.Reader, cmds []Cmder, failedCmds *cmdsMap,
 ) error {
 	for _, cmd := range cmds {
 		err := cmd.readReply(rd)
@@ -1394,7 +1381,7 @@ func (c *ClusterClient) pipelineReadCmds(
 }
 
 func (c *ClusterClient) checkMovedErr(
-	cmd Cmder, err error, failedCmds map[*clusterNode][]Cmder,
+	cmd Cmder, err error, failedCmds *cmdsMap,
 ) bool {
 	moved, ask, addr := internal.IsMovedError(err)
 
@@ -1406,7 +1393,9 @@ func (c *ClusterClient) checkMovedErr(
 			return false
 		}
 
-		failedCmds[node] = append(failedCmds[node], cmd)
+		failedCmds.mu.Lock()
+		failedCmds.m[node] = append(failedCmds.m[node], cmd)
+		failedCmds.mu.Unlock()
 		return true
 	}
 
@@ -1416,7 +1405,9 @@ func (c *ClusterClient) checkMovedErr(
 			return false
 		}
 
-		failedCmds[node] = append(failedCmds[node], NewCmd("ASKING"), cmd)
+		failedCmds.mu.Lock()
+		failedCmds.m[node] = append(failedCmds.m[node], NewCmd("ASKING"), cmd)
+		failedCmds.mu.Unlock()
 		return true
 	}
 
@@ -1456,31 +1447,38 @@ func (c *ClusterClient) defaultProcessTxPipeline(cmds []Cmder) error {
 				time.Sleep(c.retryBackoff(attempt))
 			}
 
-			failedCmds := make(map[*clusterNode][]Cmder)
+			failedCmds := newCmdsMap()
+			var wg sync.WaitGroup
 
 			for node, cmds := range cmdsMap {
-				cn, err := node.Client.getConn()
-				if err != nil {
-					if err == pool.ErrClosed {
-						c.remapCmds(cmds, failedCmds)
-					} else {
-						setCmdsErr(cmds, err)
-					}
-					continue
-				}
+				wg.Add(1)
+				go func(node *clusterNode, cmds []Cmder) {
+					defer wg.Done()
 
-				err = c.txPipelineProcessCmds(node, cn, cmds, failedCmds)
-				if err == nil || internal.IsRedisError(err) {
-					node.Client.connPool.Put(cn)
-				} else {
-					node.Client.connPool.Remove(cn)
-				}
+					cn, err := node.Client.getConn()
+					if err != nil {
+						if err == pool.ErrClosed {
+							c.mapCmdsByNode(cmds, failedCmds)
+						} else {
+							setCmdsErr(cmds, err)
+						}
+						return
+					}
+
+					err = c.txPipelineProcessCmds(node, cn, cmds, failedCmds)
+					if err == nil || internal.IsRedisError(err) {
+						node.Client.connPool.Put(cn)
+					} else {
+						node.Client.connPool.Remove(cn)
+					}
+				}(node, cmds)
 			}
 
-			if len(failedCmds) == 0 {
+			wg.Wait()
+			if len(failedCmds.m) == 0 {
 				break
 			}
-			cmdsMap = failedCmds
+			cmdsMap = failedCmds.m
 		}
 	}
 
@@ -1497,14 +1495,16 @@ func (c *ClusterClient) mapCmdsBySlot(cmds []Cmder) map[int][]Cmder {
 }
 
 func (c *ClusterClient) txPipelineProcessCmds(
-	node *clusterNode, cn *pool.Conn, cmds []Cmder, failedCmds map[*clusterNode][]Cmder,
+	node *clusterNode, cn *pool.Conn, cmds []Cmder, failedCmds *cmdsMap,
 ) error {
 	err := cn.WithWriter(c.opt.WriteTimeout, func(wr *proto.Writer) error {
 		return txPipelineWriteMulti(wr, cmds)
 	})
 	if err != nil {
 		setCmdsErr(cmds, err)
-		failedCmds[node] = cmds
+		failedCmds.mu.Lock()
+		failedCmds.m[node] = cmds
+		failedCmds.mu.Unlock()
 		return err
 	}
 
@@ -1520,7 +1520,7 @@ func (c *ClusterClient) txPipelineProcessCmds(
 }
 
 func (c *ClusterClient) txPipelineReadQueued(
-	rd *proto.Reader, cmds []Cmder, failedCmds map[*clusterNode][]Cmder,
+	rd *proto.Reader, cmds []Cmder, failedCmds *cmdsMap,
 ) error {
 	// Parse queued replies.
 	var statusCmd StatusCmd
diff --git a/ring.go b/ring.go
index 61050ca..42a4e6c 100644
--- a/ring.go
+++ b/ring.go
@@ -592,36 +592,46 @@ func (c *Ring) defaultProcessPipeline(cmds []Cmder) error {
 			time.Sleep(c.retryBackoff(attempt))
 		}
 
+		var mu sync.Mutex
 		var failedCmdsMap map[string][]Cmder
+		var wg sync.WaitGroup
 
 		for hash, cmds := range cmdsMap {
-			shard, err := c.shards.GetByHash(hash)
-			if err != nil {
-				setCmdsErr(cmds, err)
-				continue
-			}
+			wg.Add(1)
+			go func(hash string, cmds []Cmder) {
+				defer wg.Done()
 
-			cn, err := shard.Client.getConn()
-			if err != nil {
-				setCmdsErr(cmds, err)
-				continue
-			}
-
-			canRetry, err := shard.Client.pipelineProcessCmds(cn, cmds)
-			if err == nil || internal.IsRedisError(err) {
-				shard.Client.connPool.Put(cn)
-				continue
-			}
-			shard.Client.connPool.Remove(cn)
-
-			if canRetry && internal.IsRetryableError(err, true) {
-				if failedCmdsMap == nil {
-					failedCmdsMap = make(map[string][]Cmder)
+				shard, err := c.shards.GetByHash(hash)
+				if err != nil {
+					setCmdsErr(cmds, err)
+					return
 				}
-				failedCmdsMap[hash] = cmds
-			}
+
+				cn, err := shard.Client.getConn()
+				if err != nil {
+					setCmdsErr(cmds, err)
+					return
+				}
+
+				canRetry, err := shard.Client.pipelineProcessCmds(cn, cmds)
+				if err == nil || internal.IsRedisError(err) {
+					shard.Client.connPool.Put(cn)
+					return
+				}
+				shard.Client.connPool.Remove(cn)
+
+				if canRetry && internal.IsRetryableError(err, true) {
+					mu.Lock()
+					if failedCmdsMap == nil {
+						failedCmdsMap = make(map[string][]Cmder)
+					}
+					failedCmdsMap[hash] = cmds
+					mu.Unlock()
+				}
+			}(hash, cmds)
 		}
 
+		wg.Wait()
 		if len(failedCmdsMap) == 0 {
 			break
 		}