[kudu] branch master updated: [common] small optimisation on PartitionContainsRow

[alexey]

This is an automated email from the ASF dual-hosted git repository.

alexey pushed a commit to branch master
in repository https://gitbox.apache.org/repos/asf/kudu.git


The following commit(s) were added to refs/heads/master by this push:
     new cc933107d [common] small optimisation on PartitionContainsRow
cc933107d is described below

commit cc933107d9be7398b42e6c8473daea6a835ef7d7
Author: Alexey Serbin <ale...@apache.org>
AuthorDate: Thu Dec 8 19:03:56 2022 -0800

    [common] small optimisation on PartitionContainsRow
    
    This patch updates PartitionSchema::PartitionContainsRow() to avoid
    calling EncodeColumns() twice for the same data.
    
    That's to improve performance of Tablet::CheckRowInTablet() which
    is in a hot path when inserting data into a tablet.
    
    Change-Id: Iaf1d8a9ea533a859d218292a6a0e2a0818781111
    Reviewed-on: http://gerrit.cloudera.org:8080/19333
    Tested-by: Alexey Serbin <ale...@apache.org>
    Reviewed-by: Yingchun Lai <acelyc1112...@gmail.com>
---
 src/kudu/common/partition.cc | 20 ++++++++++++--------
 src/kudu/common/partition.h  |  4 ++++
 2 files changed, 16 insertions(+), 8 deletions(-)

diff --git a/src/kudu/common/partition.cc b/src/kudu/common/partition.cc
index 475d9d0ed..1669f3f5a 100644
--- a/src/kudu/common/partition.cc
+++ b/src/kudu/common/partition.cc
@@ -701,8 +701,7 @@ bool PartitionSchema::PartitionContainsRowImpl(const 
Partition& partition,
       return false;
     }
   }
-
-  return RangePartitionContainsRowImpl(partition, row);
+  return RangePartitionContainsEncodedKey(partition, range_key);
 }
 
 template<typename Row>
@@ -727,12 +726,7 @@ bool PartitionSchema::RangePartitionContainsRowImpl(
   string key;
   EncodeColumns(row, range_schema_.column_ids, &key);
 
-  // When all hash buckets match, then the row is contained in the partition
-  // if the row's key is greater or equal to the lower bound, and if there is
-  // either no upper bound or the row's key is less than the upper bound.
-  return
-      (partition.begin().range_key() <= key) &&
-      (partition.end().range_key().empty() || key < 
partition.end().range_key());
+  return RangePartitionContainsEncodedKey(partition, key);
 }
 
 bool PartitionSchema::PartitionContainsRow(const Partition& partition,
@@ -1384,6 +1378,16 @@ vector<Partition> 
PartitionSchema::GenerateHashPartitions(
   return hash_partitions;
 }
 
+bool PartitionSchema::RangePartitionContainsEncodedKey(
+    const Partition& partition, const string& key) {
+  // When all hash buckets match, then the row is contained in the partition
+  // if the row's key is greater or equal to the lower bound, and if there is
+  // either no upper bound or the row's key is less than the upper bound.
+  return
+      (partition.begin().range_key() <= key) &&
+      (partition.end().range_key().empty() || key < 
partition.end().range_key());
+}
+
 Status PartitionSchema::ValidateHashSchema(const Schema& schema,
                                            const HashSchema& hash_schema) {
   set<ColumnId> hash_columns;
diff --git a/src/kudu/common/partition.h b/src/kudu/common/partition.h
index 575c73218..20c88222e 100644
--- a/src/kudu/common/partition.h
+++ b/src/kudu/common/partition.h
@@ -564,6 +564,10 @@ class PartitionSchema {
       const HashSchema& hash_schema,
       const KeyEncoder<std::string>& hash_encoder);
 
+  // Helper for PartitionContainsRowImpl.
+  static bool RangePartitionContainsEncodedKey(const Partition& partition,
+                                               const std::string& key);
+
   // PartitionKeyDebugString implementation for row types.
   template<typename Row>
   std::string PartitionKeyDebugStringImpl(const Row& row) const;