# Frequent Reads And Writes On The Target Column  (Continue...)

+ With the gradual increase of the two parameters tidb_ddl_reorg_worker_cnt and tidb_ddl_reorg_batch_size, the impact is mainly due to write conflicts caused by concurrent ADD INDEX and Column Update. The system's performance is as follows:
  + `TiKV_prewrite_latch_wait_duration` There is a noticeable increase, causing slow writes.
  + `admin show ddl` The command can see multiple retries of the DDL job, at which point the ADD INDEX takes a very long time to complete.
------------------------------------------------------------
If the target column of the ADD INDEX is being written more frequently (this test involves UPDATE, INSERT, and DELETE of the column), the default ADD INDEX configuration has a significant impact on the online load of the system, mainly due to the concurrent execution of the ADD INDEX and the column update. The write conflict caused by the system is reflected in:
  - As the two parameters (tidb_ddl_reorg_worker_cnt and tidb_ddl_reorg_batch_size) gradually increased, TikV_Prewrite_Latch_Wait_Duration increased significantly, causing slow writes.
  - When tidb_ddl_reorg_worker_cnt and tidb_ddl_reorg_batch_size are very large, the admin show ddl command can see multiple retries of DDL jobs (for example, Write Conflict, TxnStartTS 410327455965380624 is stale [try again later], errCount:38, SnapshotVersion: 410327228136030220), at which point the ADD INDEX will take a very long time to complete.