[GitHub] [incubator-doris] morningman commented on a change in pull request #3222: Support sharding txn_map_lock into more small map locks to make good performance for txn manage task

[GitBox <gi...@apache.org>]

morningman commented on a change in pull request #3222: Support sharding txn_map_lock into more small map locks to make good performance for txn manage task
URL: https://github.com/apache/incubator-doris/pull/3222#discussion_r400933966
 
 

 ##########
 File path: be/src/olap/txn_manager.h
 ##########
 @@ -140,30 +141,48 @@ class TxnManager {
     void get_partition_ids(const TTransactionId transaction_id, std::vector<TPartitionId>* partition_ids);
     
 private:
-    RWMutex* _get_txn_lock(TTransactionId txn_id) {
-        return _txn_locks[txn_id % _txn_lock_num].get();
-    }
+
+    using TxnKey = std::pair<int64_t, int64_t>; // partition_id, transaction_id;
+
+    typedef std::map<TxnKey, std::map<TabletInfo, TabletTxnInfo>> txn_tablet_map_t;
+    typedef std::unordered_map<int64_t, std::unordered_set<int64_t>> txn_partition_map_t;
+
+    inline RWMutex& _get_txn_map_lock(TTransactionId transactionId);
+
+    inline txn_tablet_map_t& _get_txn_tablet_map(TTransactionId transactionId);
+
+    inline txn_partition_map_t& _get_txn_partition_map(TTransactionId transactionId);
 
     // insert or remove (transaction_id, partition_id) from _txn_partition_map
     // get _txn_map_lock before calling
     void _insert_txn_partition_map_unlocked(int64_t transaction_id, int64_t partition_id);
     void _clear_txn_partition_map_unlocked(int64_t transaction_id, int64_t partition_id);
 
 private:
-    RWMutex _txn_map_lock;
-    using TxnKey = std::pair<int64_t, int64_t>; // partition_id, transaction_id;
-    std::map<TxnKey, std::map<TabletInfo, TabletTxnInfo>> _txn_tablet_map;
+    const int32_t _txn_map_shard_size;
+
+    // _txn_map_locks[i] protect _txn_tablet_maps[i], i=0,1,2...,and i < _txn_map_shard_size
+    std::map<TxnKey, std::map<TabletInfo, TabletTxnInfo>> *_txn_tablet_maps;
 
 Review comment:
   And using `std::unordered_map` is better

----------------------------------------------------------------
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.
 
For queries about this service, please contact Infrastructure at:
users@infra.apache.org


With regards,
Apache Git Services

---------------------------------------------------------------------
To unsubscribe, e-mail: commits-unsubscribe@doris.apache.org
For additional commands, e-mail: commits-help@doris.apache.org