diff --git a/pkg/domain/BUILD.bazel b/pkg/domain/BUILD.bazel
index 2fdf3a0df3..d369511566 100644
--- a/pkg/domain/BUILD.bazel
+++ b/pkg/domain/BUILD.bazel
@@ -84,6 +84,7 @@ go_library(
         "//pkg/util/globalconn",
         "//pkg/util/intest",
         "//pkg/util/logutil",
+        "//pkg/util/mathutil",
         "//pkg/util/memory",
         "//pkg/util/memoryusagealarm",
         "//pkg/util/printer",
diff --git a/pkg/domain/domain.go b/pkg/domain/domain.go
index 621ded193d..239bab3ff1 100644
--- a/pkg/domain/domain.go
+++ b/pkg/domain/domain.go
@@ -87,6 +87,7 @@ import (
 	"github.com/pingcap/tidb/pkg/util/globalconn"
 	"github.com/pingcap/tidb/pkg/util/intest"
 	"github.com/pingcap/tidb/pkg/util/logutil"
+	"github.com/pingcap/tidb/pkg/util/mathutil"
 	"github.com/pingcap/tidb/pkg/util/memory"
 	"github.com/pingcap/tidb/pkg/util/memoryusagealarm"
 	"github.com/pingcap/tidb/pkg/util/replayer"
@@ -462,20 +463,22 @@ func (do *Domain) fetchAllSchemasWithTables(m *meta.Meta) ([]*model.DBInfo, erro
 const fetchSchemaConcurrency = 1
 
 func (*Domain) splitForConcurrentFetch(schemas []*model.DBInfo) [][]*model.DBInfo {
-	groupSize := (len(schemas) + fetchSchemaConcurrency - 1) / fetchSchemaConcurrency
-	if variable.SchemaCacheSize.Load() > 0 && len(schemas) > 1000 {
-		// TODO: Temporary solution to speed up when too many databases, will refactor it later.
-		groupSize = 8
-	}
-	splitted := make([][]*model.DBInfo, 0, fetchSchemaConcurrency)
+	groupCnt := fetchSchemaConcurrency
 	schemaCnt := len(schemas)
-	for i := 0; i < schemaCnt; i += groupSize {
-		end := i + groupSize
-		if end > schemaCnt {
-			end = schemaCnt
-		}
-		splitted = append(splitted, schemas[i:end])
+	if variable.SchemaCacheSize.Load() > 0 && schemaCnt > 1000 {
+		// TODO: Temporary solution to speed up when too many databases, will refactor it later.
+		groupCnt = 8
 	}
+
+	splitted := make([][]*model.DBInfo, 0, groupCnt)
+	groupSizes := mathutil.Divide2Batches(schemaCnt, groupCnt)
+
+	start := 0
+	for _, groupSize := range groupSizes {
+		splitted = append(splitted, schemas[start:start+groupSize])
+		start += groupSize
+	}
+
 	return splitted
 }