Hi all,

I would like to introduce a new feature and optimization regarding metadata 
queries. We have implemented support for the `ORDER BY TIMESERIES` clause in 
`SHOW TIMESERIES` statements. This allows users to retrieve timeseries results 
sorted lexicographically by measurement name, with support for both ascending 
and descending order.

Along with this sorting capability, we have significantly optimized the 
performance of the `OFFSET` clause. Previously, skipping records required 
traversing them one by one. With this update, we now utilize in-memory subtree 
measurement statistics to skip entire subtrees during traversal. This 
improvement drastically reduces the cost of large offsets, changing the 
complexity from linear relative to the offset to being proportional to the tree 
height.

The update also adjusts the logical plan to support global ordering across 
partitions and fixes a previous issue with limit handling in multi-partition 
queries.

You can review the changes here: https://github.com/apache/iotdb/pull/17065

Best regards,
--------------------
Yuan Tian

Reply via email to