use multi uploader

cdc/api/v2/model.go

@@ -436,6 +436,7 @@ func (c *ReplicaConfig) toInternalReplicaConfigWithOriginConfig(
 				OutputColumnID:      c.Sink.CloudStorageConfig.OutputColumnID,
 				FileExpirationDays:  c.Sink.CloudStorageConfig.FileExpirationDays,
 				FileCleanupCronSpec: c.Sink.CloudStorageConfig.FileCleanupCronSpec,
+				FlushConcurrency:    c.Sink.CloudStorageConfig.FlushConcurrency,
 			}
 		}
 
@@ -719,6 +720,7 @@ func ToAPIReplicaConfig(c *config.ReplicaConfig) *ReplicaConfig {
 				OutputColumnID:      cloned.Sink.CloudStorageConfig.OutputColumnID,
 				FileExpirationDays:  cloned.Sink.CloudStorageConfig.FileExpirationDays,
 				FileCleanupCronSpec: cloned.Sink.CloudStorageConfig.FileCleanupCronSpec,
+				FlushConcurrency:    cloned.Sink.CloudStorageConfig.FlushConcurrency,
 			}
 		}
 
@@ -1205,6 +1207,7 @@ type CloudStorageConfig struct {
 	OutputColumnID      *bool   `json:"output_column_id,omitempty"`
 	FileExpirationDays  *int    `json:"file_expiration_days,omitempty"`
 	FileCleanupCronSpec *string `json:"file_cleanup_cron_spec,omitempty"`
+	FlushConcurrency    *int    `json:"flush_concurrency,omitempty"`
 }
 
 // ChangefeedStatus holds common information of a changefeed in cdc

cdc/sink/dmlsink/cloudstorage/dml_worker.go

@@ -20,14 +20,14 @@ import (
 	"sync/atomic"
 	"time"
 
-	"github.com/pingcap/errors"
 	"github.com/pingcap/log"
 	"github.com/pingcap/tidb/br/pkg/storage"
 	"github.com/pingcap/tiflow/cdc/model"
 	"github.com/pingcap/tiflow/cdc/sink/metrics"
 	mcloudstorage "github.com/pingcap/tiflow/cdc/sink/metrics/cloudstorage"
 	"github.com/pingcap/tiflow/engine/pkg/clock"
 	"github.com/pingcap/tiflow/pkg/chann"
+	"github.com/pingcap/tiflow/pkg/errors"
 	"github.com/pingcap/tiflow/pkg/sink/cloudstorage"
 	"github.com/pingcap/tiflow/pkg/sink/codec/common"
 	"github.com/prometheus/client_golang/prometheus"
@@ -264,12 +264,35 @@ func (d *dmlWorker) writeDataFile(ctx context.Context, path string, task *single
 
 	if err := d.statistics.RecordBatchExecution(func() (int, int64, error) {
 		start := time.Now()
+		defer d.metricFlushDuration.Observe(time.Since(start).Seconds())
 
-		err := d.storage.WriteFile(ctx, path, buf.Bytes())
-		if err != nil {
-			return 0, 0, err
+		if d.config.FlushConcurrency <= 1 {
+			return rowsCnt, bytesCnt, d.storage.WriteFile(ctx, path, buf.Bytes())
+		}
+
+		writer, inErr := d.storage.Create(ctx, path, &storage.WriterOption{
+			Concurrency: d.config.FlushConcurrency,
+		})
+		if inErr != nil {
+			return 0, 0, inErr
+		}
+
+		defer func() {
+			closeErr := writer.Close(ctx)
+			if inErr != nil {
+				log.Error("failed to close writer", zap.Error(closeErr),
+					zap.Int("workerID", d.id),
+					zap.Any("table", task.tableInfo.TableName),
+					zap.String("namespace", d.changeFeedID.Namespace),
+					zap.String("changefeed", d.changeFeedID.ID))
+				if inErr == nil {
+					inErr = closeErr
+				}
+			}
+		}()
+		if _, inErr = writer.Write(ctx, buf.Bytes()); inErr != nil {
+			return 0, 0, inErr
 		}
-		d.metricFlushDuration.Observe(time.Since(start).Seconds())
 		return rowsCnt, bytesCnt, nil
 	}); err != nil {
 		return err

pkg/config/sink.go

@@ -597,6 +597,7 @@ type CloudStorageConfig struct {
 	OutputColumnID      *bool   `toml:"output-column-id" json:"output-column-id,omitempty"`
 	FileExpirationDays  *int    `toml:"file-expiration-days" json:"file-expiration-days,omitempty"`
 	FileCleanupCronSpec *string `toml:"file-cleanup-cron-spec" json:"file-cleanup-cron-spec,omitempty"`
+	FlushConcurrency    *int    `toml:"flush-concurrency" json:"flush-concurrency,omitempty"`
 }
 
 func (s *SinkConfig) validateAndAdjust(sinkURI *url.URL) error {

pkg/sink/cloudstorage/config.go

@@ -42,6 +42,12 @@ const (
 	minFlushInterval = 2 * time.Second
 	// the upper limit of flush-interval.
 	maxFlushInterval = 10 * time.Minute
+	// defaultFlushConcurrency is the default value of flush-concurrency.
+	defaultFlushConcurrency = 1
+	// the lower limit of flush-concurrency.
+	minFlushConcurrency = 1
+	// the upper limit of flush-concurrency.
+	maxFlushConcurrency = 512
 	// defaultFileSize is the default value of file-size.
 	defaultFileSize = 64 * 1024 * 1024
 	// the lower limit of file size
@@ -73,6 +79,7 @@ type Config struct {
 	FileCleanupCronSpec      string
 	EnablePartitionSeparator bool
 	OutputColumnID           bool
+	FlushConcurrency         int
 }
 
 // NewConfig returns the default cloud storage sink config.
@@ -133,11 +140,15 @@ func (c *Config) Apply(
 		if replicaConfig.Sink.CloudStorageConfig.FileCleanupCronSpec != nil {
 			c.FileCleanupCronSpec = *replicaConfig.Sink.CloudStorageConfig.FileCleanupCronSpec
 		}
+		c.FlushConcurrency = util.GetOrZero(replicaConfig.Sink.CloudStorageConfig.FlushConcurrency)
 	}
 
 	if c.FileIndexWidth < config.MinFileIndexWidth || c.FileIndexWidth > config.MaxFileIndexWidth {
 		c.FileIndexWidth = config.DefaultFileIndexWidth
 	}
+	if c.FlushConcurrency < minFlushConcurrency || c.FlushConcurrency > maxFlushConcurrency {
+		c.FlushConcurrency = defaultFlushConcurrency
+	}
 
 	return nil
 }

pkg/sink/cloudstorage/config_test.go

@@ -32,6 +32,7 @@ func TestConfigApply(t *testing.T) {
 	expected.FileIndexWidth = config.DefaultFileIndexWidth
 	expected.DateSeparator = config.DateSeparatorDay.String()
 	expected.EnablePartitionSeparator = true
+	expected.FlushConcurrency = 1
 	uri := "s3://bucket/prefix?worker-count=32&flush-interval=10s&file-size=16777216&protocol=csv"
 	sinkURI, err := url.Parse(uri)
 	require.Nil(t, err)