enhance: support configurable estimation on variable legnth field

Signed-off-by: xiaofanluan <[email protected]>

Author: xiaofan-luan

configs/milvus.yaml

@@ -846,6 +846,10 @@ common:
   defaultPartitionName: _default # Name of the default partition when a collection is created
   defaultIndexName: _default_idx # Name of the index when it is created with name unspecified
   entityExpiration: -1 # Entity expiration in seconds, CAUTION -1 means never expire
+  estimate:
+    varCharLengthLimit: 256 # Maximum length considered per VarChar value when estimating record size
+    dynamicFieldLengthLimit: 512 # Maximum length considered per JSON/Array/Geometry value when estimating record size
+    sparseFloatVectorSize: 1200 # Fallback size (bytes) for estimating SparseFloatVector fields
   indexSliceSize: 16 # Index slice size in MB
   threadCoreCoefficient:
     highPriority: 10 # This parameter specify how many times the number of threads is the number of cores in high priority pool

pkg/util/paramtable/component_param.go

@@ -32,6 +32,7 @@ import (
 	"github.com/milvus-io/milvus/pkg/v2/log"
 	"github.com/milvus-io/milvus/pkg/v2/util/hardware"
 	"github.com/milvus-io/milvus/pkg/v2/util/metricsinfo"
+	"github.com/milvus-io/milvus/pkg/v2/util/typeutil"
 )
 
 const (
@@ -269,7 +270,10 @@ type commonConfig struct {
 	TopicNames             ParamItem `refreshable:"true"`
 	TimeTicker             ParamItem `refreshable:"true"`
 
-	JSONMaxLength ParamItem `refreshable:"false"`
+	JSONMaxLength                 ParamItem `refreshable:"false"`
+	DynamicFieldMaxLength         ParamItem `refreshable:"true"`
+	SparseFloatVectorEstimateSize ParamItem `refreshable:"true"`
+	VarCharEstimateLengthLimit    ParamItem `refreshable:"true"`
 
 	MetricsPort ParamItem `refreshable:"false"`
 
@@ -875,6 +879,45 @@ Large numeric passwords require double quotes to avoid yaml parsing precision is
 	}
 	p.JSONMaxLength.Init(base.mgr)
 
+	p.DynamicFieldMaxLength = ParamItem{
+		Key:          "common.estimate.dynamicFieldLengthLimit",
+		Version:      "2.5.0",
+		DefaultValue: fmt.Sprint(typeutil.GetDynamicFieldEstimateLength()),
+		Doc:          "maximum length used when estimating JSON/Array/Geometry field size",
+		Export:       true,
+	}
+	p.DynamicFieldMaxLength.Init(base.mgr)
+	typeutil.SetDynamicFieldEstimateLength(p.DynamicFieldMaxLength.GetAsInt())
+	base.mgr.Dispatcher.Register(p.DynamicFieldMaxLength.Key, config.NewHandler(p.DynamicFieldMaxLength.Key, func(event *config.Event) {
+		typeutil.SetDynamicFieldEstimateLength(p.DynamicFieldMaxLength.GetAsInt())
+	}))
+
+	p.SparseFloatVectorEstimateSize = ParamItem{
+		Key:          "common.estimate.sparseFloatVectorSize",
+		Version:      "2.5.0",
+		DefaultValue: fmt.Sprint(typeutil.GetSparseFloatVectorEstimateLength()),
+		Doc:          "fallback size (bytes) used when estimating sparse float vector fields",
+		Export:       true,
+	}
+	p.SparseFloatVectorEstimateSize.Init(base.mgr)
+	typeutil.SetSparseFloatVectorEstimateLength(p.SparseFloatVectorEstimateSize.GetAsInt())
+	base.mgr.Dispatcher.Register(p.SparseFloatVectorEstimateSize.Key, config.NewHandler(p.SparseFloatVectorEstimateSize.Key, func(event *config.Event) {
+		typeutil.SetSparseFloatVectorEstimateLength(p.SparseFloatVectorEstimateSize.GetAsInt())
+	}))
+
+	p.VarCharEstimateLengthLimit = ParamItem{
+		Key:          "common.estimate.varCharLengthLimit",
+		Version:      "2.5.0",
+		DefaultValue: fmt.Sprint(typeutil.GetVarCharEstimateLength()),
+		Doc:          "maximum length used when estimating VarChar field size",
+		Export:       true,
+	}
+	p.VarCharEstimateLengthLimit.Init(base.mgr)
+	typeutil.SetVarCharEstimateLength(p.VarCharEstimateLengthLimit.GetAsInt())
+	base.mgr.Dispatcher.Register(p.VarCharEstimateLengthLimit.Key, config.NewHandler(p.VarCharEstimateLengthLimit.Key, func(event *config.Event) {
+		typeutil.SetVarCharEstimateLength(p.VarCharEstimateLengthLimit.GetAsInt())
+	}))
+
 	p.MetricsPort = ParamItem{
 		Key:          "common.MetricsPort",
 		Version:      "2.3.0",

pkg/util/typeutil/schema.go

@@ -37,8 +37,6 @@ import (
 	"github.com/milvus-io/milvus/pkg/v2/log"
 )
 
-const DynamicFieldMaxLength = 512
-
 type getVariableFieldLengthPolicy int
 
 const (
@@ -75,16 +73,17 @@ func getVarFieldLength(fieldSchema *schemapb.FieldSchema, policy getVariableFiel
 			// TODO this is a hack and may not accurate, we should rely on estimate size per record
 			// However we should report size and datacoord calculate based on size
 			// https://github.com/milvus-io/milvus/issues/17687
-			if maxLength > 256 {
-				return 256, nil
+			estimateLimit := GetVarCharEstimateLength()
+			if maxLength > estimateLimit {
+				return estimateLimit, nil
 			}
 			return maxLength, nil
 		default:
 			return 0, fmt.Errorf("unrecognized getVariableFieldLengthPolicy %v", policy)
 		}
-	// geometry field max length now consider the same as json field, which is 512 bytes
+		// geometry field max length now consider the same as json field, which is 512 bytes
 	case schemapb.DataType_Array, schemapb.DataType_JSON, schemapb.DataType_Geometry:
-		return DynamicFieldMaxLength, nil
+		return GetDynamicFieldEstimateLength(), nil
 	default:
 		return 0, fmt.Errorf("field %s is not a variable-length type", fieldSchema.DataType.String())
 	}
@@ -160,7 +159,7 @@ func estimateSizeBy(schema *schemapb.CollectionSchema, policy getVariableFieldLe
 			// varies depending on the number of non-zeros. Using sparse vector
 			// generated by SPLADE as reference and returning size of a sparse
 			// vector with 150 non-zeros.
-			res += 1200
+			res += GetSparseFloatVectorEstimateLength()
 		}
 	}
 	return res, nil

pkg/util/typeutil/schema_test.go

@@ -177,11 +177,81 @@ func TestSchema(t *testing.T) {
 	}
 
 	t.Run("EstimateSizePerRecord", func(t *testing.T) {
+		limit := GetDynamicFieldEstimateLength()
 		size, err := EstimateSizePerRecord(schema)
-		assert.Equal(t, 680+DynamicFieldMaxLength*4, size)
+		assert.Equal(t, 680+limit*4, size)
 		assert.NoError(t, err)
 	})
 
+	t.Run("VarCharEstimateLengthLimit", func(t *testing.T) {
+		originalLimit := GetVarCharEstimateLength()
+		t.Cleanup(func() { SetVarCharEstimateLength(originalLimit) })
+
+		field := &schemapb.FieldSchema{
+			DataType: schemapb.DataType_VarChar,
+			TypeParams: []*commonpb.KeyValuePair{
+				{
+					Key:   common.MaxLengthKey,
+					Value: "1024",
+				},
+			},
+		}
+
+		SetVarCharEstimateLength(128)
+		length, err := getVarFieldLength(field, custom)
+		assert.NoError(t, err)
+		assert.Equal(t, 128, length)
+
+		SetVarCharEstimateLength(4096)
+		length, err = getVarFieldLength(field, custom)
+		assert.NoError(t, err)
+		assert.Equal(t, 1024, length)
+	})
+
+	t.Run("DynamicFieldMaxLengthLimit", func(t *testing.T) {
+		originalLimit := GetDynamicFieldEstimateLength()
+		t.Cleanup(func() { SetDynamicFieldEstimateLength(originalLimit) })
+
+		field := &schemapb.FieldSchema{
+			DataType: schemapb.DataType_JSON,
+		}
+
+		SetDynamicFieldEstimateLength(2048)
+		length, err := getVarFieldLength(field, custom)
+		assert.NoError(t, err)
+		assert.Equal(t, 2048, length)
+
+		SetDynamicFieldEstimateLength(128)
+		length, err = getVarFieldLength(field, custom)
+		assert.NoError(t, err)
+		assert.Equal(t, 128, length)
+	})
+
+	t.Run("SparseFloatVectorEstimateSize", func(t *testing.T) {
+		original := GetSparseFloatVectorEstimateLength()
+		t.Cleanup(func() { SetSparseFloatVectorEstimateLength(original) })
+
+		schemaWithSparse := &schemapb.CollectionSchema{
+			Fields: []*schemapb.FieldSchema{
+				{
+					FieldID:  1200,
+					Name:     "sparse",
+					DataType: schemapb.DataType_SparseFloatVector,
+				},
+			},
+		}
+
+		SetSparseFloatVectorEstimateLength(2048)
+		size, err := EstimateSizePerRecord(schemaWithSparse)
+		assert.NoError(t, err)
+		assert.Equal(t, 2048, size)
+
+		SetSparseFloatVectorEstimateLength(64)
+		size, err = EstimateSizePerRecord(schemaWithSparse)
+		assert.NoError(t, err)
+		assert.Equal(t, 64, size)
+	})
+
 	t.Run("SchemaHelper", func(t *testing.T) {
 		_, err := CreateSchemaHelper(nil)
 		assert.Error(t, err)