fix

jerry-024 · jerry-024 · commit 12ad08f93dab · 2026-02-12T19:14:16.000+08:00
diff --git a/paimon-diskann/PARAMETER_TUNING.md b/paimon-diskann/PARAMETER_TUNING.md
@@ -108,14 +108,6 @@ DiskANN is a graph-based approximate nearest neighbor (ANN) search algorithm des
   - **2,000,000**: Default for balanced performance
   - **5,000,000+**: For large-scale production systems with ample resources
 
-#### `vector.diskann.index-type`
-- **Default**: MEMORY
-- **Options**: MEMORY, DISK
-- **Description**: Type of index structure
-- **Recommendations**:
-  - **MEMORY**: For datasets that fit in RAM (best performance)
-  - **DISK**: For datasets exceeding RAM (requires SSD)
-
 ## Performance Tuning Guide
 
 ### High Recall (>95%)
@@ -139,14 +131,6 @@ vector.diskann.build-list-size = 75
 vector.diskann.search-list-size = 32
 ```
 
-### Memory-Constrained
-```properties
-vector.diskann.max-degree = 32
-vector.diskann.build-list-size = 75
-vector.size-per-index = 500000
-vector.diskann.index-type = DISK
-```
-
 ## Best Practices
 
 1. **Start with defaults**: The default parameters are tuned for balanced performance
diff --git a/paimon-diskann/paimon-diskann-e2e-test/src/test/scala/org/apache/paimon/spark/sql/DiskAnnVectorIndexE2ETest.scala b/paimon-diskann/paimon-diskann-e2e-test/src/test/scala/org/apache/paimon/spark/sql/DiskAnnVectorIndexE2ETest.scala
@@ -82,7 +82,7 @@ class DiskAnnVectorIndexE2ETest extends PaimonSparkTestBase {
       spark.sql(s"INSERT INTO T VALUES $values")
 
       val output = spark
-        .sql("CALL sys.create_global_index(table => 'test.T', index_column => 'v', index_type => 'diskann-vector-ann', options => 'vector.dim=3,vector.diskann.index-type=MEMORY')")
+        .sql("CALL sys.create_global_index(table => 'test.T', index_column => 'v', index_type => 'diskann-vector-ann', options => 'vector.dim=3')")
         .collect()
         .head
       assert(output.getBoolean(0))
diff --git a/paimon-diskann/paimon-diskann-index/src/main/java/org/apache/paimon/diskann/index/DiskAnnIndex.java b/paimon-diskann/paimon-diskann-index/src/main/java/org/apache/paimon/diskann/index/DiskAnnIndex.java
@@ -45,14 +45,9 @@ private DiskAnnIndex(Index index, int dimension, int buildListSize) {
     }
 
     public static DiskAnnIndex create(
-            int dimension,
-            DiskAnnVectorMetric metric,
-            DiskAnnIndexType indexType,
-            int maxDegree,
-            int buildListSize) {
+            int dimension, DiskAnnVectorMetric metric, int maxDegree, int buildListSize) {
         MetricType metricType = metric.toMetricType();
-        Index index =
-                Index.create(dimension, metricType, indexType.value(), maxDegree, buildListSize);
+        Index index = Index.create(dimension, metricType, 0, maxDegree, buildListSize);
         return new DiskAnnIndex(index, dimension, buildListSize);
     }
 
diff --git a/paimon-diskann/paimon-diskann-index/src/main/java/org/apache/paimon/diskann/index/DiskAnnIndexType.java b/paimon-diskann/paimon-diskann-index/src/main/java/org/apache/paimon/diskann/index/DiskAnnIndexType.java
diff --git a/paimon-diskann/paimon-diskann-index/src/main/java/org/apache/paimon/diskann/index/DiskAnnVectorGlobalIndexWriter.java b/paimon-diskann/paimon-diskann-index/src/main/java/org/apache/paimon/diskann/index/DiskAnnVectorGlobalIndexWriter.java
@@ -278,7 +278,7 @@ private void flushCurrentIndex() throws IOException {
                 new DiskAnnIndexMeta(
                         dim,
                         options.metric().toMetricType().value(),
-                        options.indexType().value(),
+                        0,
                         currentIndexCount,
                         currentIndexMinId,
                         currentIndexMaxId,
@@ -300,7 +300,6 @@ private DiskAnnIndex createIndex() {
         return DiskAnnIndex.create(
                 options.dimension(),
                 options.metric(),
-                options.indexType(),
                 options.maxDegree(),
                 options.buildListSize());
     }
diff --git a/paimon-diskann/paimon-diskann-index/src/main/java/org/apache/paimon/diskann/index/DiskAnnVectorIndexOptions.java b/paimon-diskann/paimon-diskann-index/src/main/java/org/apache/paimon/diskann/index/DiskAnnVectorIndexOptions.java
@@ -38,12 +38,6 @@ public class DiskAnnVectorIndexOptions {
                     .withDescription(
                             "The similarity metric for vector search (L2, INNER_PRODUCT, COSINE), and L2 is the default");
 
-    public static final ConfigOption<DiskAnnIndexType> VECTOR_INDEX_TYPE =
-            ConfigOptions.key("vector.diskann.index-type")
-                    .enumType(DiskAnnIndexType.class)
-                    .defaultValue(DiskAnnIndexType.MEMORY)
-                    .withDescription("The DiskANN index type to use (MEMORY, DISK)");
-
     public static final ConfigOption<Integer> VECTOR_MAX_DEGREE =
             ConfigOptions.key("vector.diskann.max-degree")
                     .intType()
@@ -99,7 +93,6 @@ public class DiskAnnVectorIndexOptions {
 
     private final int dimension;
     private final DiskAnnVectorMetric metric;
-    private final DiskAnnIndexType indexType;
     private final int maxDegree;
     private final int buildListSize;
     private final int searchListSize;
@@ -112,7 +105,6 @@ public class DiskAnnVectorIndexOptions {
     public DiskAnnVectorIndexOptions(Options options) {
         this.dimension = options.get(VECTOR_DIM);
         this.metric = options.get(VECTOR_METRIC);
-        this.indexType = options.get(VECTOR_INDEX_TYPE);
         this.maxDegree = options.get(VECTOR_MAX_DEGREE);
         this.buildListSize = options.get(VECTOR_BUILD_LIST_SIZE);
         this.searchListSize = options.get(VECTOR_SEARCH_LIST_SIZE);
@@ -137,10 +129,6 @@ public DiskAnnVectorMetric metric() {
         return metric;
     }
 
-    public DiskAnnIndexType indexType() {
-        return indexType;
-    }
-
     public int maxDegree() {
         return maxDegree;
     }
diff --git a/paimon-diskann/paimon-diskann-index/src/main/java/org/apache/paimon/diskann/index/FileIOGraphReader.java b/paimon-diskann/paimon-diskann-index/src/main/java/org/apache/paimon/diskann/index/FileIOGraphReader.java
@@ -70,7 +70,6 @@ public class FileIOGraphReader implements Closeable {
     // ---- Header fields ----
     private final int dimension;
     private final int metricValue;
-    private final int indexTypeValue;
     private final int maxDegree;
     private final int buildListSize;
     private final int count;
@@ -113,7 +112,7 @@ public FileIOGraphReader(SeekableInputStream input, int cacheSize) throws IOExce
         off += 4;
         this.metricValue = readInt(headerBuf, off);
         off += 4;
-        this.indexTypeValue = readInt(headerBuf, off);
+        // skip indexType(4)
         off += 4;
         this.maxDegree = readInt(headerBuf, off);
         off += 4;
@@ -172,10 +171,6 @@ public int getMetricValue() {
         return metricValue;
     }
 
-    public int getIndexTypeValue() {
-        return indexTypeValue;
-    }
-
     public int getMaxDegree() {
         return maxDegree;
     }
diff --git a/paimon-diskann/paimon-diskann-index/src/test/java/org/apache/paimon/diskann/index/DiskAnnVectorGlobalIndexScanTest.java b/paimon-diskann/paimon-diskann-index/src/test/java/org/apache/paimon/diskann/index/DiskAnnVectorGlobalIndexScanTest.java
@@ -105,7 +105,6 @@ public void before() throws Exception {
                         .option(CoreOptions.BUCKET.key(), "-1")
                         .option("vector.dim", "2")
                         .option("vector.metric", "L2")
-                        .option("vector.diskann.index-type", "MEMORY")
                         .option("data-evolution.enabled", "true")
                         .option("row-tracking.enabled", "true")
                         .build();
@@ -160,7 +159,6 @@ public void testVectorIndexScanWithDifferentMetrics() throws Exception {
                         .option(CoreOptions.BUCKET.key(), "-1")
                         .option("vector.dim", "2")
                         .option("vector.metric", "INNER_PRODUCT")
-                        .option("vector.diskann.index-type", "MEMORY")
                         .option("data-evolution.enabled", "true")
                         .option("row-tracking.enabled", "true")
                         .build();
diff --git a/paimon-diskann/paimon-diskann-index/src/test/java/org/apache/paimon/diskann/index/DiskAnnVectorGlobalIndexTest.java b/paimon-diskann/paimon-diskann-index/src/test/java/org/apache/paimon/diskann/index/DiskAnnVectorGlobalIndexTest.java
@@ -153,43 +153,36 @@ public void testDifferentMetrics() throws IOException {
     }
 
     @Test
-    public void testDifferentIndexTypes() throws IOException {
+    public void testDefaultOptions() throws IOException {
         int dimension = 32;
         int numVectors = 100;
 
-        String[] indexTypes = {"MEMORY"};
-
-        for (String indexType : indexTypes) {
-            Options options = createDefaultOptions(dimension);
-            options.setString("vector.diskann.index-type", indexType);
-            DiskAnnVectorIndexOptions indexOptions = new DiskAnnVectorIndexOptions(options);
-            Path typeIndexPath = new Path(indexPath, indexType.toLowerCase());
-            GlobalIndexFileWriter fileWriter = createFileWriter(typeIndexPath);
-            DiskAnnVectorGlobalIndexWriter writer =
-                    new DiskAnnVectorGlobalIndexWriter(fileWriter, vectorType, indexOptions);
+        Options options = createDefaultOptions(dimension);
+        DiskAnnVectorIndexOptions indexOptions = new DiskAnnVectorIndexOptions(options);
+        GlobalIndexFileWriter fileWriter = createFileWriter(indexPath);
+        DiskAnnVectorGlobalIndexWriter writer =
+                new DiskAnnVectorGlobalIndexWriter(fileWriter, vectorType, indexOptions);
 
-            List<float[]> testVectors = generateRandomVectors(numVectors, dimension);
-            testVectors.forEach(writer::write);
+        List<float[]> testVectors = generateRandomVectors(numVectors, dimension);
+        testVectors.forEach(writer::write);
 
-            List<ResultEntry> results = writer.finish();
-            assertThat(results).hasSize(1);
+        List<ResultEntry> results = writer.finish();
+        assertThat(results).hasSize(1);
 
-            ResultEntry result = results.get(0);
-            GlobalIndexFileReader fileReader = createFileReader(typeIndexPath);
-            List<GlobalIndexIOMeta> metas = new ArrayList<>();
-            metas.add(
-                    new GlobalIndexIOMeta(
-                            new Path(typeIndexPath, result.fileName()),
-                            fileIO.getFileSize(new Path(typeIndexPath, result.fileName())),
-                            result.meta()));
+        ResultEntry result = results.get(0);
+        GlobalIndexFileReader fileReader = createFileReader(indexPath);
+        List<GlobalIndexIOMeta> metas = new ArrayList<>();
+        metas.add(
+                new GlobalIndexIOMeta(
+                        new Path(indexPath, result.fileName()),
+                        fileIO.getFileSize(new Path(indexPath, result.fileName())),
+                        result.meta()));
 
-            try (DiskAnnVectorGlobalIndexReader reader =
-                    new DiskAnnVectorGlobalIndexReader(
-                            fileReader, metas, vectorType, indexOptions)) {
-                VectorSearch vectorSearch = new VectorSearch(testVectors.get(0), 5, fieldName);
-                GlobalIndexResult searchResult = reader.visitVectorSearch(vectorSearch).get();
-                assertThat(searchResult).isNotNull();
-            }
+        try (DiskAnnVectorGlobalIndexReader reader =
+                new DiskAnnVectorGlobalIndexReader(fileReader, metas, vectorType, indexOptions)) {
+            VectorSearch vectorSearch = new VectorSearch(testVectors.get(0), 5, fieldName);
+            GlobalIndexResult searchResult = reader.visitVectorSearch(vectorSearch).get();
+            assertThat(searchResult).isNotNull();
         }
     }
 
@@ -448,7 +441,6 @@ private Options createDefaultOptions(int dimension) {
         Options options = new Options();
         options.setInteger("vector.dim", dimension);
         options.setString("vector.metric", "L2");
-        options.setString("vector.diskann.index-type", "MEMORY");
         options.setInteger("vector.diskann.max-degree", 64);
         options.setInteger("vector.diskann.build-list-size", 100);
         options.setInteger("vector.diskann.search-list-size", 100);
diff --git a/paimon-diskann/paimon-diskann-jni/src/main/native/src/lib.rs b/paimon-diskann/paimon-diskann-jni/src/main/native/src/lib.rs
@@ -1055,7 +1055,6 @@ pub extern "system" fn Java_org_apache_paimon_diskann_DiskAnnNative_indexCreateS
 
     let dimension   = call_int!("getDimension");
     let metric_type = call_int!("getMetricValue");
-    let _index_type = call_int!("getIndexTypeValue");
     let max_degree  = call_int!("getMaxDegree") as usize;
     let build_ls    = call_int!("getBuildListSize") as usize;
     let count       = call_int!("getCount") as usize;
