You are viewing a plain text version of this content. The canonical link for it is here.
Posted to gitbox@hive.apache.org by "dengzhhu653 (via GitHub)" <gi...@apache.org> on 2023/02/02 05:26:27 UTC
[GitHub] [hive] dengzhhu653 commented on a diff in pull request #3905: HIVE-26035: Implement direct SQL for add partitions to improve performance at HMS
dengzhhu653 commented on code in PR #3905:
URL: https://github.com/apache/hive/pull/3905#discussion_r1094044633
##########
standalone-metastore/metastore-server/src/main/java/org/apache/hadoop/hive/metastore/DirectSqlInsertPart.java:
##########
@@ -0,0 +1,804 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package org.apache.hadoop.hive.metastore;
+
+import static org.apache.commons.lang3.StringUtils.repeat;
+import static org.apache.hadoop.hive.metastore.Batchable.NO_BATCHING;
+
+import java.util.ArrayList;
+import java.util.HashMap;
+import java.util.Iterator;
+import java.util.List;
+import java.util.Map;
+
+import javax.jdo.PersistenceManager;
+
+import org.apache.hadoop.hive.metastore.api.MetaException;
+import org.apache.hadoop.hive.metastore.model.MColumnDescriptor;
+import org.apache.hadoop.hive.metastore.model.MFieldSchema;
+import org.apache.hadoop.hive.metastore.model.MOrder;
+import org.apache.hadoop.hive.metastore.model.MPartition;
+import org.apache.hadoop.hive.metastore.model.MPartitionColumnPrivilege;
+import org.apache.hadoop.hive.metastore.model.MPartitionPrivilege;
+import org.apache.hadoop.hive.metastore.model.MSerDeInfo;
+import org.apache.hadoop.hive.metastore.model.MStorageDescriptor;
+import org.apache.hadoop.hive.metastore.model.MStringList;
+import org.datanucleus.ExecutionContext;
+import org.datanucleus.api.jdo.JDOPersistenceManager;
+import org.datanucleus.identity.DatastoreId;
+import org.datanucleus.metadata.AbstractClassMetaData;
+import org.datanucleus.metadata.IdentityType;
+
+/**
+ * This class contains the methods to insert into tables on the underlying database using direct SQL
+ */
+class DirectSqlInsertPart {
+ private final PersistenceManager pm;
+ private final DatabaseProduct dbType;
+ private final int batchSize;
+
+ public DirectSqlInsertPart(PersistenceManager pm, DatabaseProduct dbType, int batchSize) {
+ this.pm = pm;
+ this.dbType = dbType;
+ this.batchSize = batchSize;
+ }
+
+ /**
+ * Interface to execute multiple row insert query in batch for direct SQL
+ */
+ interface BatchExecutionContext {
+ void execute(String batchQueryText, int batchRowCount, int batchParamCount) throws MetaException;
+ }
+
+ private Long getDataStoreId(Class<?> modelClass) throws MetaException {
+ ExecutionContext ec = ((JDOPersistenceManager) pm).getExecutionContext();
+ AbstractClassMetaData cmd = ec.getMetaDataManager().getMetaDataForClass(modelClass, ec.getClassLoaderResolver());
+ if (cmd.getIdentityType() == IdentityType.DATASTORE) {
+ return (Long) ec.getStoreManager().getValueGenerationStrategyValue(ec, cmd, -1);
+ } else {
+ throw new MetaException("Identity type is not datastore.");
+ }
+ }
+
+ private void insertInBatch(String tableName, String columns, int columnCount, int rowCount,
+ BatchExecutionContext batchExecutionContext) throws MetaException {
+ if (rowCount == 0 || columnCount == 0) {
+ return;
+ }
+ int maxRowsInBatch = (batchSize == NO_BATCHING) ? rowCount : batchSize;
+ int maxBatches = rowCount / maxRowsInBatch;
+ int last = rowCount % maxRowsInBatch;
+ String rowFormat = "(" + repeat(",?", columnCount).substring(1) + ")";
+ String query = "";
+ if (maxBatches > 0) {
+ query = dbType.getBatchInsertQuery(tableName, columns, rowFormat, maxRowsInBatch);
+ }
+ int batchParamCount = maxRowsInBatch * columnCount;
+ for (int batch = 0; batch < maxBatches; batch++) {
+ batchExecutionContext.execute(query, maxRowsInBatch, batchParamCount);
+ }
+ if (last != 0) {
+ query = dbType.getBatchInsertQuery(tableName, columns, rowFormat, last);
+ batchExecutionContext.execute(query, last, last * columnCount);
+ }
+ }
+
+ private void executeQuery(String queryText, Object[] params) throws MetaException {
+ try (QueryWrapper query = new QueryWrapper(pm.newQuery("javax.jdo.query.SQL", queryText))) {
+ MetastoreDirectSqlUtils.executeWithArray(query.getInnerQuery(), params, queryText);
+ }
+ }
+
+ private void insertSerdeInBatch(Map<Long, MSerDeInfo> serdeIdToSerDeInfo) throws MetaException {
+ int rowCount = serdeIdToSerDeInfo.size();
+ String columns = "(\"SERDE_ID\",\"DESCRIPTION\",\"DESERIALIZER_CLASS\",\"NAME\",\"SERDE_TYPE\",\"SLIB\","
+ + "\"SERIALIZER_CLASS\")";
+ int columnCount = 7;
+ BatchExecutionContext batchExecutionContext = new BatchExecutionContext() {
+ final Iterator<Map.Entry<Long, MSerDeInfo>> it = serdeIdToSerDeInfo.entrySet().iterator();
+ @Override
+ public void execute(String batchQueryText, int batchRowCount, int batchParamCount) throws MetaException {
+ Object[] params = new Object[batchParamCount];
+ int paramIndex = 0;
+ for (int index = 0; index < batchRowCount; index++) {
+ Map.Entry<Long, MSerDeInfo> entry = it.next();
+ MSerDeInfo serdeInfo = entry.getValue();
+ params[paramIndex++] = entry.getKey();
+ params[paramIndex++] = serdeInfo.getDescription();
+ params[paramIndex++] = serdeInfo.getDeserializerClass();
+ params[paramIndex++] = serdeInfo.getName();
+ params[paramIndex++] = serdeInfo.getSerdeType();
+ params[paramIndex++] = serdeInfo.getSerializationLib();
+ params[paramIndex++] = serdeInfo.getSerializerClass();
+ }
+ executeQuery(batchQueryText, params);
+ }
+ };
+ insertInBatch("\"SERDES\"", columns, columnCount, rowCount, batchExecutionContext);
+ }
+
+ private void insertStorageDescriptorInBatch(Map<Long, MStorageDescriptor> sdIdToStorageDescriptor,
+ Map<Long, Long> sdIdToSerdeId, Map<Long, Long> sdIdToCdId) throws MetaException {
+ int rowCount = sdIdToStorageDescriptor.size();
+ String columns = "(\"SD_ID\",\"CD_ID\",\"INPUT_FORMAT\",\"IS_COMPRESSED\",\"IS_STOREDASSUBDIRECTORIES\","
+ + "\"LOCATION\",\"NUM_BUCKETS\",\"OUTPUT_FORMAT\",\"SERDE_ID\")";
+ int columnCount = 9;
+ BatchExecutionContext batchExecutionContext = new BatchExecutionContext() {
+ final Iterator<Map.Entry<Long, MStorageDescriptor>> it = sdIdToStorageDescriptor.entrySet().iterator();
+ @Override
+ public void execute(String batchQueryText, int batchRowCount, int batchParamCount) throws MetaException {
+ Object[] params = new Object[batchParamCount];
+ int paramIndex = 0;
+ for (int index = 0; index < batchRowCount; index++) {
+ Map.Entry<Long, MStorageDescriptor> entry = it.next();
+ MStorageDescriptor sd = entry.getValue();
+ params[paramIndex++] = entry.getKey();
+ params[paramIndex++] = sdIdToCdId.get(entry.getKey());
+ params[paramIndex++] = sd.getInputFormat();
+ params[paramIndex++] = dbType.getBoolean(sd.isCompressed());
+ params[paramIndex++] = dbType.getBoolean(sd.isStoredAsSubDirectories());
+ params[paramIndex++] = sd.getLocation();
+ params[paramIndex++] = sd.getNumBuckets();
+ params[paramIndex++] = sd.getOutputFormat();
+ params[paramIndex++] = sdIdToSerdeId.get(entry.getKey());
+ }
+ executeQuery(batchQueryText, params);
+ }
+ };
+ insertInBatch("\"SDS\"", columns, columnCount, rowCount, batchExecutionContext);
+ }
+
+ private void insertPartitionInBatch(Map<Long, MPartition> partIdToPartition, Map<Long, Long> partIdToSdId)
+ throws MetaException {
+ int rowCount = partIdToPartition.size();
+ String columns = "(\"PART_ID\",\"CREATE_TIME\",\"LAST_ACCESS_TIME\",\"PART_NAME\",\"SD_ID\",\"TBL_ID\","
+ + "\"WRITE_ID\")";
+ int columnCount = 7;
+ BatchExecutionContext batchExecutionContext = new BatchExecutionContext() {
+ final Iterator<Map.Entry<Long, MPartition>> it = partIdToPartition.entrySet().iterator();
+ @Override
+ public void execute(String batchQueryText, int batchRowCount, int batchParamCount) throws MetaException {
+ Object[] params = new Object[batchParamCount];
+ int paramIndex = 0;
+ for (int index = 0; index < batchRowCount; index++) {
+ Map.Entry<Long, MPartition> entry = it.next();
+ MPartition partition = entry.getValue();
+ params[paramIndex++] = entry.getKey();
+ params[paramIndex++] = partition.getCreateTime();
+ params[paramIndex++] = partition.getLastAccessTime();
+ params[paramIndex++] = partition.getPartitionName();
+ params[paramIndex++] = partIdToSdId.get(entry.getKey());
+ params[paramIndex++] = partition.getTable().getId();
+ params[paramIndex++] = partition.getWriteId();
+ }
+ executeQuery(batchQueryText, params);
+ }
+ };
+ insertInBatch("\"PARTITIONS\"", columns, columnCount, rowCount, batchExecutionContext);
+ }
+
+ private void insertSerdeParamInBatch(Map<Long, MSerDeInfo> serdeIdToSerDeInfo) throws MetaException {
+ int rowCount = 0;
+ for (MSerDeInfo serDeInfo : serdeIdToSerDeInfo.values()) {
+ rowCount += serDeInfo.getParameters().size();
+ }
+ if (rowCount == 0) {
+ return;
+ }
+ String columns = "(\"SERDE_ID\",\"PARAM_KEY\",\"PARAM_VALUE\")";
+ int columnCount = 3;
+ BatchExecutionContext batchExecutionContext = new BatchExecutionContext() {
+ final Iterator<Map.Entry<Long, MSerDeInfo>> serdeIt = serdeIdToSerDeInfo.entrySet().iterator();
+ Map.Entry<Long, MSerDeInfo> serdeEntry = serdeIt.next();
+ Iterator<Map.Entry<String, String>> it = serdeEntry.getValue().getParameters().entrySet().iterator();
+ @Override
+ public void execute(String batchQueryText, int batchRowCount, int batchParamCount) throws MetaException {
+ Object[] params = new Object[batchParamCount];
+ int index = 0;
+ int paramIndex = 0;
+ do {
+ while (index < batchRowCount && it.hasNext()) {
+ Map.Entry<String, String> entry = it.next();
+ params[paramIndex++] = serdeEntry.getKey();
+ params[paramIndex++] = entry.getKey();
+ params[paramIndex++] = entry.getValue();
+ index++;
+ }
+ if (index < batchRowCount) {
+ serdeEntry = serdeIt.next();
Review Comment:
Nit: the `skewedColNames`, `skewedColValues` and `skewedColValueLocationMaps` may be null in MStorageDescriptor:
https://github.com/apache/hive/blob/master/standalone-metastore/metastore-server/src/main/java/org/apache/hadoop/hive/metastore/ObjectStore.java#L2526-L2542
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: gitbox-unsubscribe@hive.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: gitbox-unsubscribe@hive.apache.org
For additional commands, e-mail: gitbox-help@hive.apache.org