Houliang Qi created IOTDB-1150:
----------------------------------
Summary: Support concurrent asynchronous execution of
CreateMultiTimeSeriesPlan in distribuetd version
Key: IOTDB-1150
URL: https://issues.apache.org/jira/browse/IOTDB-1150
Project: Apache IoTDB
Issue Type: Improvement
Components: Cluster
Reporter: Houliang Qi
When using the _createMultiTimeseries()_ interface to create time series, a
CreateMultiTimeSeriesPlan plan will be generated. In order to simplify the lock
competition of the same data raft group, we split the CreateMultiTimeSeriesPlan
plan into several sub CreateMultiTimeSeriesPlans according to different data
raft groups, and each sub CreateMultiTimeSeriesPlans belongs to the same data
raft group. However, we did not consider a problem, that is, these sub
CreateMultiTimeSeriesPlans may belong to different storage groups. When
applying these plans, we can not use the advantage of asynchronous concurrent
applier, which makes it appropriate. As long as we encounter the
CreateMultiTimeSeriesPlan plan, all operations are serial applied.
Therefore, it needs to be optimized. The optimization method is also very
simple. When splitting the plan, you can split it more finely. For timeseries
that belongs to the same data group and the same SG, assemble one
CreateMultiTimeSeriesPlan. In this way, when applying the plan, different SG of
the same data group can be applied in parallel. The implementation of this part
of code can refer to the split of InsertMultiTabletPlan.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
