enhance: autoindex for multi data type

Signed-off-by: chasingegg <chao.gao@zilliz.com>

internal/proxy/task_index.go

@@ -263,10 +263,23 @@ func (cit *createIndexTask) parseIndexParams() error {
 				return err
 			}
 		}
-		if indexType == indexparamcheck.IndexSparseInverted || indexType == indexparamcheck.IndexSparseWand {
+		if isVecIndex {
 			metricType, metricTypeExist := indexParamsMap[common.MetricTypeKey]
-			if !metricTypeExist || metricType != metric.IP {
-				return fmt.Errorf("only IP is the supported metric type for sparse index")
+			if !metricTypeExist {
+				return fmt.Errorf("metric type not set for vector index")
+			}
+			if typeutil.IsDenseFloatVectorType(cit.fieldSchema.DataType) {
+				if !funcutil.SliceContain(indexparamcheck.FloatVectorMetrics, metricType) {
+					return fmt.Errorf("float vector index does not support metric type: " + metricType)
+				}
+			} else if typeutil.IsSparseFloatVectorType(cit.fieldSchema.DataType) {
+				if metricType != metric.IP {
+					return fmt.Errorf("only IP is the supported metric type for sparse index")
+				}
+			} else if typeutil.IsBinaryVectorType(cit.fieldSchema.DataType) {
+				if !funcutil.SliceContain(indexparamcheck.BinaryVectorMetrics, metricType) {
+					return fmt.Errorf("binary vector index does not support metric type: " + metricType)
+				}
 			}
 		}
 	}

internal/querynodev2/optimizers/query_hook.go

@@ -62,6 +62,7 @@ func OptimizeSearchParams(ctx context.Context, req *querypb.SearchRequest, query
 			common.SearchParamKey:  queryInfo.GetSearchParams(),
 			common.SegmentNumKey:   estSegmentNum,
 			common.WithFilterKey:   withFilter,
+			common.DataTypeKey:     plan.GetVectorAnns().GetVectorType(),
 			common.WithOptimizeKey: paramtable.Get().AutoIndexConfig.EnableOptimize.GetAsBool(),
 			common.CollectionKey:   req.GetReq().GetCollectionID(),
 		}

pkg/common/common.go

@@ -99,6 +99,7 @@ const (
 	SearchParamKey  = "search_param"
 	SegmentNumKey   = "segment_num"
 	WithFilterKey   = "with_filter"
+	DataTypeKey     = "data_type"
 	WithOptimizeKey = "with_optimize"
 	CollectionKey   = "collection"
 

pkg/util/indexparamcheck/base_checker.go

@@ -59,7 +59,7 @@ func (c baseChecker) CheckValidDataType(dType schemapb.DataType) error {
 	return nil
 }
 
-func (c baseChecker) SetDefaultMetricTypeIfNotExist(m map[string]string) {}
+func (c baseChecker) SetDefaultMetricTypeIfNotExist(m map[string]string, dType schemapb.DataType) {}
 
 func (c baseChecker) StaticCheck(params map[string]string) error {
 	return errors.New("unsupported index type")

pkg/util/indexparamcheck/binary_vector_base_checker.go

@@ -34,7 +34,7 @@ func (c binaryVectorBaseChecker) CheckValidDataType(dType schemapb.DataType) err
 	return nil
 }
 
-func (c binaryVectorBaseChecker) SetDefaultMetricTypeIfNotExist(params map[string]string) {
+func (c binaryVectorBaseChecker) SetDefaultMetricTypeIfNotExist(params map[string]string, dType schemapb.DataType) {
 	setDefaultIfNotExist(params, common.MetricTypeKey, BinaryVectorDefaultMetricType)
 }
 

pkg/util/indexparamcheck/constraints.go

@@ -46,8 +46,10 @@ const (
 	SparseDropRatioBuild = "drop_ratio_build"
 )
 
-// METRICS is a set of all metrics types supported for float vector.
-var METRICS = []string{metric.L2, metric.IP, metric.COSINE} // const
+var (
+	FloatVectorMetrics  = []string{metric.L2, metric.IP, metric.COSINE}                                        // const
+	BinaryVectorMetrics = []string{metric.HAMMING, metric.JACCARD, metric.SUBSTRUCTURE, metric.SUPERSTRUCTURE} // const
+)
 
 // BinIDMapMetrics is a set of all metric types supported for binary vector.
 var (

pkg/util/indexparamcheck/float_vector_base_checker.go

@@ -13,8 +13,8 @@ type floatVectorBaseChecker struct {
 }
 
 func (c floatVectorBaseChecker) staticCheck(params map[string]string) error {
-	if !CheckStrByValues(params, Metric, METRICS) {
-		return fmt.Errorf("metric type %s not found or not supported, supported: %v", params[Metric], METRICS)
+	if !CheckStrByValues(params, Metric, FloatVectorMetrics) {
+		return fmt.Errorf("metric type %s not found or not supported, supported: %v", params[Metric], FloatVectorMetrics)
 	}
 
 	return nil
@@ -35,7 +35,7 @@ func (c floatVectorBaseChecker) CheckValidDataType(dType schemapb.DataType) erro
 	return nil
 }
 
-func (c floatVectorBaseChecker) SetDefaultMetricTypeIfNotExist(params map[string]string) {
+func (c floatVectorBaseChecker) SetDefaultMetricTypeIfNotExist(params map[string]string, dType schemapb.DataType) {
 	setDefaultIfNotExist(params, common.MetricTypeKey, FloatVectorDefaultMetricType)
 }
 

pkg/util/indexparamcheck/hnsw_checker.go

@@ -4,11 +4,12 @@ import (
 	"fmt"
 
 	"github.com/milvus-io/milvus-proto/go-api/v2/schemapb"
+	"github.com/milvus-io/milvus/pkg/common"
 	"github.com/milvus-io/milvus/pkg/util/typeutil"
 )
 
 type hnswChecker struct {
-	floatVectorBaseChecker
+	baseChecker
 }
 
 func (c hnswChecker) StaticCheck(params map[string]string) error {
@@ -38,6 +39,16 @@ func (c hnswChecker) CheckValidDataType(dType schemapb.DataType) error {
 	return nil
 }
 
+func (c hnswChecker) SetDefaultMetricTypeIfNotExist(params map[string]string, dType schemapb.DataType) {
+	if typeutil.IsDenseFloatVectorType(dType) {
+		setDefaultIfNotExist(params, common.MetricTypeKey, FloatVectorDefaultMetricType)
+	} else if typeutil.IsSparseFloatVectorType(dType) {
+		setDefaultIfNotExist(params, common.MetricTypeKey, SparseFloatVectorDefaultMetricType)
+	} else if typeutil.IsBinaryVectorType(dType) {
+		setDefaultIfNotExist(params, common.MetricTypeKey, BinaryVectorDefaultMetricType)
+	}
+}
+
 func newHnswChecker() IndexChecker {
 	return &hnswChecker{}
 }

pkg/util/indexparamcheck/hnsw_checker_test.go

@@ -172,3 +172,42 @@ func Test_hnswChecker_CheckValidDataType(t *testing.T) {
 		}
 	}
 }
+
+func Test_hnswChecker_SetDefaultMetricType(t *testing.T) {
+	cases := []struct {
+		dType      schemapb.DataType
+		metricType string
+	}{
+		{
+			dType:      schemapb.DataType_FloatVector,
+			metricType: metric.IP,
+		},
+		{
+			dType:      schemapb.DataType_Float16Vector,
+			metricType: metric.IP,
+		},
+		{
+			dType:      schemapb.DataType_BFloat16Vector,
+			metricType: metric.IP,
+		},
+		{
+			dType:      schemapb.DataType_SparseFloatVector,
+			metricType: metric.IP,
+		},
+		{
+			dType:      schemapb.DataType_BinaryVector,
+			metricType: metric.JACCARD,
+		},
+	}
+
+	c := newHnswChecker()
+	for _, test := range cases {
+		p := map[string]string{
+			DIM:            strconv.Itoa(128),
+			HNSWM:          strconv.Itoa(16),
+			EFConstruction: strconv.Itoa(200),
+		}
+		c.SetDefaultMetricTypeIfNotExist(p, test.dType)
+		assert.Equal(t, p[Metric], test.metricType)
+	}
+}

pkg/util/indexparamcheck/index_checker.go

@@ -23,6 +23,6 @@ import (
 type IndexChecker interface {
 	CheckTrain(map[string]string) error
 	CheckValidDataType(dType schemapb.DataType) error
-	SetDefaultMetricTypeIfNotExist(map[string]string)
+	SetDefaultMetricTypeIfNotExist(map[string]string, schemapb.DataType)
 	StaticCheck(map[string]string) error
 }

pkg/util/indexparamcheck/sparse_float_vector_base_checker.go

@@ -39,7 +39,7 @@ func (c sparseFloatVectorBaseChecker) CheckValidDataType(dType schemapb.DataType
 	return nil
 }
 
-func (c sparseFloatVectorBaseChecker) SetDefaultMetricTypeIfNotExist(params map[string]string) {
+func (c sparseFloatVectorBaseChecker) SetDefaultMetricTypeIfNotExist(params map[string]string, dType schemapb.DataType) {
 	setDefaultIfNotExist(params, common.MetricTypeKey, SparseFloatVectorDefaultMetricType)
 }
 

pkg/util/paramtable/autoindex_param.go

@@ -19,6 +19,7 @@ package paramtable
 import (
 	"fmt"
 
+	"github.com/milvus-io/milvus-proto/go-api/v2/schemapb"
 	"github.com/milvus-io/milvus/pkg/common"
 	"github.com/milvus-io/milvus/pkg/config"
 	"github.com/milvus-io/milvus/pkg/util/funcutil"
@@ -193,31 +194,37 @@ func (p *autoIndexConfig) init(base *BaseTable) {
 }
 
 func (p *autoIndexConfig) panicIfNotValidAndSetDefaultMetricType(mgr *config.Manager) {
-	m := p.IndexParams.GetAsJSONMap()
+	p.panicIfNotValidAndSetDefaultMetricTypeHelper(p.IndexParams.Key, p.IndexParams.GetAsJSONMap(), schemapb.DataType_FloatVector, mgr)
+	p.panicIfNotValidAndSetDefaultMetricTypeHelper(p.BinaryIndexParams.Key, p.BinaryIndexParams.GetAsJSONMap(), schemapb.DataType_BinaryVector, mgr)
+	p.panicIfNotValidAndSetDefaultMetricTypeHelper(p.SparseIndexParams.Key, p.SparseIndexParams.GetAsJSONMap(), schemapb.DataType_SparseFloatVector, mgr)
+}
+
+func (p *autoIndexConfig) panicIfNotValidAndSetDefaultMetricTypeHelper(key string, m map[string]string, dtype schemapb.DataType, mgr *config.Manager) {
+	// handle := func(key string, m map[string]string, mgr *config.Manager, dtype schemapb.DataType) {
 	if m == nil {
-		panic("autoIndex.build not invalid, should be json format")
+		panic(fmt.Sprintf("%s invalid, should be json format", key))
 	}
 
 	indexType, ok := m[common.IndexTypeKey]
 	if !ok {
-		panic("autoIndex.build not invalid, index type not found")
+		panic(fmt.Sprintf("%s invalid, index type not found", key))
 	}
 
 	checker, err := indexparamcheck.GetIndexCheckerMgrInstance().GetChecker(indexType)
 	if err != nil {
-		panic(fmt.Sprintf("autoIndex.build not invalid, unsupported index type: %s", indexType))
+		panic(fmt.Sprintf("%s invalid, unsupported index type: %s", key, indexType))
 	}
 
-	checker.SetDefaultMetricTypeIfNotExist(m)
+	checker.SetDefaultMetricTypeIfNotExist(m, dtype)
 
 	if err := checker.StaticCheck(m); err != nil {
-		panic(fmt.Sprintf("autoIndex.build not invalid, parameters not invalid, error: %s", err.Error()))
+		panic(fmt.Sprintf("%s invalid, parameters invalid, error: %s", key, err.Error()))
 	}
 
-	p.reset(m, mgr)
+	p.reset(key, m, mgr)
 }
 
-func (p *autoIndexConfig) reset(m map[string]string, mgr *config.Manager) {
+func (p *autoIndexConfig) reset(key string, m map[string]string, mgr *config.Manager) {
 	j := funcutil.MapToJSON(m)
-	mgr.SetConfig("autoIndex.params.build", string(j))
+	mgr.SetConfig(key, string(j))
 }

pkg/util/paramtable/autoindex_param_test.go

@@ -23,6 +23,7 @@ import (
 
 	"github.com/stretchr/testify/assert"
 
+	"github.com/milvus-io/milvus-proto/go-api/v2/schemapb"
 	"github.com/milvus-io/milvus/pkg/common"
 	"github.com/milvus-io/milvus/pkg/config"
 	"github.com/milvus-io/milvus/pkg/util/indexparamcheck"
@@ -140,7 +141,7 @@ func Test_autoIndexConfig_panicIfNotValid(t *testing.T) {
 		}
 		p.IndexParams.Init(mgr)
 		assert.Panics(t, func() {
-			p.panicIfNotValidAndSetDefaultMetricType(mgr)
+			p.panicIfNotValidAndSetDefaultMetricTypeHelper(p.IndexParams.Key, p.IndexParams.GetAsJSONMap(), schemapb.DataType_FloatVector, mgr)
 		})
 	})
 
@@ -154,7 +155,7 @@ func Test_autoIndexConfig_panicIfNotValid(t *testing.T) {
 		}
 		p.IndexParams.Init(mgr)
 		assert.Panics(t, func() {
-			p.panicIfNotValidAndSetDefaultMetricType(mgr)
+			p.panicIfNotValidAndSetDefaultMetricTypeHelper(p.IndexParams.Key, p.IndexParams.GetAsJSONMap(), schemapb.DataType_FloatVector, mgr)
 		})
 	})
 
@@ -168,7 +169,7 @@ func Test_autoIndexConfig_panicIfNotValid(t *testing.T) {
 		}
 		p.IndexParams.Init(mgr)
 		assert.Panics(t, func() {
-			p.panicIfNotValidAndSetDefaultMetricType(mgr)
+			p.panicIfNotValidAndSetDefaultMetricTypeHelper(p.IndexParams.Key, p.IndexParams.GetAsJSONMap(), schemapb.DataType_FloatVector, mgr)
 		})
 	})
 
@@ -182,13 +183,47 @@ func Test_autoIndexConfig_panicIfNotValid(t *testing.T) {
 		}
 		p.IndexParams.Init(mgr)
 		assert.NotPanics(t, func() {
-			p.panicIfNotValidAndSetDefaultMetricType(mgr)
+			p.panicIfNotValidAndSetDefaultMetricTypeHelper(p.IndexParams.Key, p.IndexParams.GetAsJSONMap(), schemapb.DataType_FloatVector, mgr)
 		})
 		metricType, exist := p.IndexParams.GetAsJSONMap()[common.MetricTypeKey]
 		assert.True(t, exist)
 		assert.Equal(t, indexparamcheck.FloatVectorDefaultMetricType, metricType)
 	})
 
+	t.Run("normal case, binary vector", func(t *testing.T) {
+		mgr := config.NewManager()
+		mgr.SetConfig("autoIndex.params.binary.build", `{"nlist": 1024, "index_type": "BIN_IVF_FLAT"}`)
+		p := &autoIndexConfig{
+			BinaryIndexParams: ParamItem{
+				Key: "autoIndex.params.binary.build",
+			},
+		}
+		p.BinaryIndexParams.Init(mgr)
+		assert.NotPanics(t, func() {
+			p.panicIfNotValidAndSetDefaultMetricTypeHelper(p.BinaryIndexParams.Key, p.BinaryIndexParams.GetAsJSONMap(), schemapb.DataType_BinaryVector, mgr)
+		})
+		metricType, exist := p.BinaryIndexParams.GetAsJSONMap()[common.MetricTypeKey]
+		assert.True(t, exist)
+		assert.Equal(t, indexparamcheck.BinaryVectorDefaultMetricType, metricType)
+	})
+
+	t.Run("normal case, sparse vector", func(t *testing.T) {
+		mgr := config.NewManager()
+		mgr.SetConfig("autoIndex.params.sparse.build", `{"index_type": "SPARSE_INVERTED_INDEX", "metric_type": "IP"}`)
+		p := &autoIndexConfig{
+			SparseIndexParams: ParamItem{
+				Key: "autoIndex.params.sparse.build",
+			},
+		}
+		p.SparseIndexParams.Init(mgr)
+		assert.NotPanics(t, func() {
+			p.panicIfNotValidAndSetDefaultMetricTypeHelper(p.SparseIndexParams.Key, p.SparseIndexParams.GetAsJSONMap(), schemapb.DataType_SparseFloatVector, mgr)
+		})
+		metricType, exist := p.SparseIndexParams.GetAsJSONMap()[common.MetricTypeKey]
+		assert.True(t, exist)
+		assert.Equal(t, indexparamcheck.SparseFloatVectorDefaultMetricType, metricType)
+	})
+
 	t.Run("normal case, ivf flat", func(t *testing.T) {
 		mgr := config.NewManager()
 		mgr.SetConfig("autoIndex.params.build", `{"nlist": 30, "index_type": "IVF_FLAT"}`)
@@ -199,7 +234,7 @@ func Test_autoIndexConfig_panicIfNotValid(t *testing.T) {
 		}
 		p.IndexParams.Init(mgr)
 		assert.NotPanics(t, func() {
-			p.panicIfNotValidAndSetDefaultMetricType(mgr)
+			p.panicIfNotValidAndSetDefaultMetricTypeHelper(p.IndexParams.Key, p.IndexParams.GetAsJSONMap(), schemapb.DataType_FloatVector, mgr)
 		})
 		metricType, exist := p.IndexParams.GetAsJSONMap()[common.MetricTypeKey]
 		assert.True(t, exist)
@@ -216,7 +251,7 @@ func Test_autoIndexConfig_panicIfNotValid(t *testing.T) {
 		}
 		p.IndexParams.Init(mgr)
 		assert.NotPanics(t, func() {
-			p.panicIfNotValidAndSetDefaultMetricType(mgr)
+			p.panicIfNotValidAndSetDefaultMetricTypeHelper(p.IndexParams.Key, p.IndexParams.GetAsJSONMap(), schemapb.DataType_FloatVector, mgr)
 		})
 		metricType, exist := p.IndexParams.GetAsJSONMap()[common.MetricTypeKey]
 		assert.True(t, exist)
@@ -233,7 +268,7 @@ func Test_autoIndexConfig_panicIfNotValid(t *testing.T) {
 		}
 		p.IndexParams.Init(mgr)
 		assert.NotPanics(t, func() {
-			p.panicIfNotValidAndSetDefaultMetricType(mgr)
+			p.panicIfNotValidAndSetDefaultMetricTypeHelper(p.IndexParams.Key, p.IndexParams.GetAsJSONMap(), schemapb.DataType_FloatVector, mgr)
 		})
 		metricType, exist := p.IndexParams.GetAsJSONMap()[common.MetricTypeKey]
 		assert.True(t, exist)
@@ -250,7 +285,7 @@ func Test_autoIndexConfig_panicIfNotValid(t *testing.T) {
 		}
 		p.IndexParams.Init(mgr)
 		assert.NotPanics(t, func() {
-			p.panicIfNotValidAndSetDefaultMetricType(mgr)
+			p.panicIfNotValidAndSetDefaultMetricTypeHelper(p.IndexParams.Key, p.IndexParams.GetAsJSONMap(), schemapb.DataType_FloatVector, mgr)
 		})
 		metricType, exist := p.IndexParams.GetAsJSONMap()[common.MetricTypeKey]
 		assert.True(t, exist)
@@ -267,7 +302,7 @@ func Test_autoIndexConfig_panicIfNotValid(t *testing.T) {
 		}
 		p.IndexParams.Init(mgr)
 		assert.NotPanics(t, func() {
-			p.panicIfNotValidAndSetDefaultMetricType(mgr)
+			p.panicIfNotValidAndSetDefaultMetricTypeHelper(p.IndexParams.Key, p.IndexParams.GetAsJSONMap(), schemapb.DataType_FloatVector, mgr)
 		})
 		metricType, exist := p.IndexParams.GetAsJSONMap()[common.MetricTypeKey]
 		assert.True(t, exist)