virajjasani opened a new pull request, #2297:
URL: https://github.com/apache/phoenix/pull/2297

   Jira: PHOENIX-7709
   
   In case of the mutable covered index, the sequence of mutations that takes 
place synchronously are
   
   1. Unverified index row mutations
   2. Data table row updates
   3. Verified index row mutations
   
   Phoenix provides config 
"org.apache.hadoop.hbase.index.lazy.post_batch.write" to enable async or lazy 
mode for the last phase i.e. verified index row mutations.
   
   If the config is enabled, it uses LazyParallelWriterIndexCommitter 
implementation which prepares the batch of mutations per table and prepares the 
TaskBatch objects. However, it does not submit the mutations to the thread-pool 
as of today. As a result, if the lazy mode is enabled, every read on the index 
table will end up performing read-repair.
   
   The purpose of this Jira is to submit the post writer tasks to the 
respective thread-pool without getting blocked on it, to be able to make the 
third phase asynchronous. When the number of indexes are higher, this mode can 
help reduce the write latencies of the data table updates and improve the 
overall throughput.
   
   User can enable this mode for the given table by using DDL statements.
   
   e.g.
   - `ALTER TABLE T1 SET 
"org.apache.hadoop.hbase.index.lazy.post_batch.write"=true`
   - `CREATE TABLE T1 (PK1 VARCHAR NOT NULL, C1 VARCHAR) 
"org.apache.hadoop.hbase.index.lazy.post_batch.write"=true`


-- 
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: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]

Reply via email to