[GitHub] drill pull request #638: DRILL-4982: Separate Hive reader classes for differ...

[paul-rogers]

Github user paul-rogers commented on a diff in the pull request:

https://github.com/apache/drill/pull/638#discussion_r91446474
  
--- Diff: 
contrib/storage-hive/core/src/main/codegen/templates/HiveRecordReaders.java ---
@@ -0,0 +1,300 @@
+/**
+ * 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.
+ */
+
+/**
+ * This template is used to generate different Hive record reader classes 
for different data formats
+ * to avoid JIT profile pullusion. These readers are derived from 
HiveAbstractReader which implements
+ * codes for init and setup stage, but the repeated - and performance 
critical part - next() method is
+ * separately implemented in the classes generated from this template. The 
internal SkipRecordReeader
+ * class is also separated as well due to the same reason.
+ *
+ * As to the performance gain with this change, please refer to:
+ * https://issues.apache.org/jira/browse/DRILL-4982
+ *
+ */
+<@pp.dropOutputFile />
+<#list hiveFormat.map as entry>
+<@pp.changeOutputFile 
name="/org/apache/drill/exec/store/hive/Hive${entry.hiveReader}Reader.java" />
+<#include "/@includes/license.ftl" />
+
+package org.apache.drill.exec.store.hive;
+
+import java.io.IOException;
+import java.util.List;
+import java.util.Properties;
+import org.apache.drill.common.exceptions.DrillRuntimeException;
+import org.apache.drill.common.exceptions.ExecutionSetupException;
+import org.apache.drill.common.expression.SchemaPath;
+import org.apache.drill.exec.ops.FragmentContext;
+import org.apache.drill.exec.vector.AllocationHelper;
+import org.apache.drill.exec.vector.ValueVector;
+import org.apache.hadoop.hive.metastore.api.Partition;
+import org.apache.hadoop.hive.metastore.api.Table;
+import org.apache.hadoop.io.Writable;
+import org.apache.hadoop.mapred.InputSplit;
+import org.apache.hadoop.security.UserGroupInformation;
+import org.apache.hadoop.hive.conf.HiveConf;
+
+import org.apache.hadoop.hive.serde2.SerDeException;
+
+import org.apache.hadoop.mapred.RecordReader;
+<#if entry.hasHeaderFooter == true>
+import org.apache.hadoop.hive.serde2.SerDe;
+import org.apache.hadoop.hive.serde2.objectinspector.StructObjectInspector;
+import com.google.common.collect.Lists;
+import java.util.ArrayList;
+import java.util.Arrays;
+import java.util.HashSet;
+import java.util.Queue;
+import java.util.Set;
+import org.apache.hadoop.hive.metastore.api.hive_metastoreConstants;
+import org.apache.hadoop.hive.serde.serdeConstants;
+
+
+public class Hive${entry.hiveReader}Reader extends HiveAbstractReader {
+
+  Object key;
+<#if entry.hasHeaderFooter == true>
--- End diff --

Free marker allows simple boolean expressions:
`<#if entry.hasHeaderFooter>`


---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastruct...@apache.org or file a JIRA ticket
with INFRA.
---

[GitHub] drill pull request #638: DRILL-4982: Separate Hive reader classes for differ...

[paul-rogers]

Github user paul-rogers commented on a diff in the pull request:

https://github.com/apache/drill/pull/638#discussion_r91448234
  
--- Diff: contrib/storage-hive/core/src/main/codegen/data/HiveFormats.tdd 
---
@@ -0,0 +1,50 @@
+# 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.
+
+{
--- End diff --

Can we explain this a bit? We have 6 reader types. But, the only difference 
in generated code is has header/footer or not. Can we solve the Java 
optimization problem with a "classic" type hierarchy:

```
HiveAbstractReader
. HiveSimpleReader
. . HiveAvroReader
. . ...
. HiveHeaderFooterReader
. . HiveTextReader
. . ...
```

Names are just made up. The point is, can a much simpler Java hierarchy, 
with less duplicated code, solve the problem? If there is one function that is 
sub-optimized, can just that one function be generated in the subclass rather 
than generating duplicate code?


---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastruct...@apache.org or file a JIRA ticket
with INFRA.
---

[GitHub] drill pull request #638: DRILL-4982: Separate Hive reader classes for differ...

[paul-rogers]

Github user paul-rogers commented on a diff in the pull request:

https://github.com/apache/drill/pull/638#discussion_r91447926
  
--- Diff: 
contrib/storage-hive/core/src/main/codegen/templates/HiveRecordReaders.java ---
@@ -0,0 +1,300 @@
+/**
+ * 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.
+ */
+
+/**
+ * This template is used to generate different Hive record reader classes 
for different data formats
+ * to avoid JIT profile pullusion. These readers are derived from 
HiveAbstractReader which implements
+ * codes for init and setup stage, but the repeated - and performance 
critical part - next() method is
+ * separately implemented in the classes generated from this template. The 
internal SkipRecordReeader
+ * class is also separated as well due to the same reason.
+ *
+ * As to the performance gain with this change, please refer to:
+ * https://issues.apache.org/jira/browse/DRILL-4982
+ *
+ */
+<@pp.dropOutputFile />
+<#list hiveFormat.map as entry>
+<@pp.changeOutputFile 
name="/org/apache/drill/exec/store/hive/Hive${entry.hiveReader}Reader.java" />
+<#include "/@includes/license.ftl" />
+
+package org.apache.drill.exec.store.hive;
+
+import java.io.IOException;
+import java.util.List;
+import java.util.Properties;
+import org.apache.drill.common.exceptions.DrillRuntimeException;
+import org.apache.drill.common.exceptions.ExecutionSetupException;
+import org.apache.drill.common.expression.SchemaPath;
+import org.apache.drill.exec.ops.FragmentContext;
+import org.apache.drill.exec.vector.AllocationHelper;
+import org.apache.drill.exec.vector.ValueVector;
+import org.apache.hadoop.hive.metastore.api.Partition;
+import org.apache.hadoop.hive.metastore.api.Table;
+import org.apache.hadoop.io.Writable;
+import org.apache.hadoop.mapred.InputSplit;
+import org.apache.hadoop.security.UserGroupInformation;
+import org.apache.hadoop.hive.conf.HiveConf;
+
+import org.apache.hadoop.hive.serde2.SerDeException;
+
+import org.apache.hadoop.mapred.RecordReader;
+<#if entry.hasHeaderFooter == true>
+import org.apache.hadoop.hive.serde2.SerDe;
+import org.apache.hadoop.hive.serde2.objectinspector.StructObjectInspector;
+import com.google.common.collect.Lists;
+import java.util.ArrayList;
+import java.util.Arrays;
+import java.util.HashSet;
+import java.util.Queue;
+import java.util.Set;
+import org.apache.hadoop.hive.metastore.api.hive_metastoreConstants;
+import org.apache.hadoop.hive.serde.serdeConstants;
+
+
+public class Hive${entry.hiveReader}Reader extends HiveAbstractReader {
+
+  Object key;
+<#if entry.hasHeaderFooter == true>
+  SkipRecordsInspector skipRecordsInspector;
+<#else>
+  Object value;
+
+
+  public Hive${entry.hiveReader}Reader(Table table, Partition partition, 
InputSplit inputSplit, List projectedColumns,
+   FragmentContext context, final HiveConf hiveConf,
+   UserGroupInformation proxyUgi) throws 
ExecutionSetupException {
+super(table, partition, inputSplit, projectedColumns, context, 
hiveConf, proxyUgi);
+  }
+
+  public  void internalInit(Properties tableProperties, 
RecordReader reader) {
+
+key = reader.createKey();
+<#if entry.hasHeaderFooter == true>
+skipRecordsInspector = new SkipRecordsInspector(tableProperties, 
reader);
+<#else>
+value = reader.createValue();
+
+
+  }
+  private void readHiveRecordAndInsertIntoRecordBatch(Object 
deSerializedValue, int outputRecordIndex) {
+for (int i = 0; i < selectedStructFieldRefs.size(); i++) {
+  Object hiveValue = finalOI.getStructFieldData(deSerializedValue, 
selectedStructFieldRefs.get(i));
+  if (hiveValue != null) {
+
selectedColumnFieldConverters.get(i).setSafeValue(selectedColumnObjInspectors.get(i),
 hiveValue,
+  vectors.get(i), outputRecordIndex);
+  }
+}
+  }

[GitHub] drill pull request #638: DRILL-4982: Separate Hive reader classes for differ...

[paul-rogers]

Github user paul-rogers commented on a diff in the pull request:

https://github.com/apache/drill/pull/638#discussion_r91447651
  
--- Diff: 
contrib/storage-hive/core/src/main/codegen/templates/HiveRecordReaders.java ---
@@ -0,0 +1,300 @@
+/**
+ * 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.
+ */
+
+/**
+ * This template is used to generate different Hive record reader classes 
for different data formats
+ * to avoid JIT profile pullusion. These readers are derived from 
HiveAbstractReader which implements
+ * codes for init and setup stage, but the repeated - and performance 
critical part - next() method is
+ * separately implemented in the classes generated from this template. The 
internal SkipRecordReeader
+ * class is also separated as well due to the same reason.
+ *
+ * As to the performance gain with this change, please refer to:
+ * https://issues.apache.org/jira/browse/DRILL-4982
+ *
+ */
+<@pp.dropOutputFile />
+<#list hiveFormat.map as entry>
+<@pp.changeOutputFile 
name="/org/apache/drill/exec/store/hive/Hive${entry.hiveReader}Reader.java" />
+<#include "/@includes/license.ftl" />
+
+package org.apache.drill.exec.store.hive;
+
+import java.io.IOException;
+import java.util.List;
+import java.util.Properties;
+import org.apache.drill.common.exceptions.DrillRuntimeException;
+import org.apache.drill.common.exceptions.ExecutionSetupException;
+import org.apache.drill.common.expression.SchemaPath;
+import org.apache.drill.exec.ops.FragmentContext;
+import org.apache.drill.exec.vector.AllocationHelper;
+import org.apache.drill.exec.vector.ValueVector;
+import org.apache.hadoop.hive.metastore.api.Partition;
+import org.apache.hadoop.hive.metastore.api.Table;
+import org.apache.hadoop.io.Writable;
+import org.apache.hadoop.mapred.InputSplit;
+import org.apache.hadoop.security.UserGroupInformation;
+import org.apache.hadoop.hive.conf.HiveConf;
+
+import org.apache.hadoop.hive.serde2.SerDeException;
+
+import org.apache.hadoop.mapred.RecordReader;
+<#if entry.hasHeaderFooter == true>
+import org.apache.hadoop.hive.serde2.SerDe;
+import org.apache.hadoop.hive.serde2.objectinspector.StructObjectInspector;
+import com.google.common.collect.Lists;
+import java.util.ArrayList;
+import java.util.Arrays;
+import java.util.HashSet;
+import java.util.Queue;
+import java.util.Set;
+import org.apache.hadoop.hive.metastore.api.hive_metastoreConstants;
+import org.apache.hadoop.hive.serde.serdeConstants;
+
+
+public class Hive${entry.hiveReader}Reader extends HiveAbstractReader {
+
+  Object key;
+<#if entry.hasHeaderFooter == true>
+  SkipRecordsInspector skipRecordsInspector;
+<#else>
+  Object value;
+
+
+  public Hive${entry.hiveReader}Reader(Table table, Partition partition, 
InputSplit inputSplit, List projectedColumns,
+   FragmentContext context, final HiveConf hiveConf,
+   UserGroupInformation proxyUgi) throws 
ExecutionSetupException {
+super(table, partition, inputSplit, projectedColumns, context, 
hiveConf, proxyUgi);
+  }
+
+  public  void internalInit(Properties tableProperties, 
RecordReader reader) {
+
+key = reader.createKey();
+<#if entry.hasHeaderFooter == true>
+skipRecordsInspector = new SkipRecordsInspector(tableProperties, 
reader);
+<#else>
+value = reader.createValue();
+
+
+  }
+  private void readHiveRecordAndInsertIntoRecordBatch(Object 
deSerializedValue, int outputRecordIndex) {
+for (int i = 0; i < selectedStructFieldRefs.size(); i++) {
+  Object hiveValue = finalOI.getStructFieldData(deSerializedValue, 
selectedStructFieldRefs.get(i));
+  if (hiveValue != null) {
+
selectedColumnFieldConverters.get(i).setSafeValue(selectedColumnObjInspectors.get(i),
 hiveValue,
+  vectors.get(i), outputRecordIndex);
+  }
+}
+  }

[GitHub] drill pull request #638: DRILL-4982: Separate Hive reader classes for differ...

[asfgit]

Github user asfgit closed the pull request at:

https://github.com/apache/drill/pull/638


---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastruct...@apache.org or file a JIRA ticket
with INFRA.
---

[GitHub] drill pull request #638: DRILL-4982: Separate Hive reader classes for differ...

[bitblender]

Github user bitblender commented on a diff in the pull request:

https://github.com/apache/drill/pull/638#discussion_r90574793
  
--- Diff: 
contrib/storage-hive/core/src/main/java/org/apache/drill/exec/store/hive/HiveAbstractReader.java
 ---
@@ -218,17 +229,18 @@ private void init() throws ExecutionSetupException {
 }
   }
 } catch (Exception e) {
-  throw new ExecutionSetupException("Failure while initializing 
HiveRecordReader: " + e.getMessage(), e);
+  throw new ExecutionSetupException("Failure while initializing 
HiveOrcReader: " + e.getMessage(), e);
--- End diff --

"HiveOrcReader:" in the exception-string should be changed so that it 
prints the actual reader type. 


---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastruct...@apache.org or file a JIRA ticket
with INFRA.
---

[GitHub] drill pull request #638: DRILL-4982: Separate Hive reader classes for differ...

[chunhui-shi]

GitHub user chunhui-shi opened a pull request:

https://github.com/apache/drill/pull/638

DRILL-4982: Separate Hive reader classes for different data formats t…

…o improve performance.

1, Separating Hive reader classes allows optimization to apply on different 
classes in optimized ways. This  separation effectively avoid the performance 
degradation of scan.

2, Do not apply Skip footer/header mechanism on most Hive formats. This 
skip mechanism introduces extra checks on each incoming records.

You can merge this pull request into a Git repository by running:

$ git pull https://github.com/chunhui-shi/drill DRILL-4982

Alternatively you can review and apply these changes as the patch at:

https://github.com/apache/drill/pull/638.patch

To close this pull request, make a commit to your master/trunk branch
with (at least) the following in the commit message:

This closes #638


commit 1e240a0dfea7970798bd550204388fc0e9bc3d42
Author: chunhui-shi 
Date:   2016-10-30T08:29:06Z

DRILL-4982: Separate Hive reader classes for different data formats to 
improve performance.

1, Separating Hive reader classes allows optimization to apply on different 
classes in optimized ways. This  separation effectively avoid the performance 
degradation of scan.

2, Do not apply Skip footer/header mechanism on most Hive formats. This 
skip mechanism introduces extra checks on each incoming records.




---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastruct...@apache.org or file a JIRA ticket
with INFRA.
---