AggSpillManager is used to manage how to write spill data file and read them back.
The spill data first need to be partitioned. Each partition contains 2 files: agg_state_file and input_chunks_file.
The spill file consume a data chunk and serialize the chunk into a protobuf bytes.
Finally, spill file content will look like the below.
The file write pattern is append-only and the read pattern is sequential scan.
This can maximize the disk IO performance.
DeleteExecutor implements table deletion with values from its child executor.
Distributed Lookup Join Executor.
High level Execution flow:
Repeat 1-3:
Group Top-N Executor
HashAggExecutor implements the hash aggregate algorithm.
Hash Join Executor
InsertExecutor implements table insertion with values from its child executor.
Limit executor.
Local Lookup Join Executor.
High level Execution flow:
Repeat 1-3:
Lookup Join Base.
Used by LocalLookupJoinExecutor and DistributedLookupJoinExecutor.
ManagedExecutor build on top of the underlying executor. For now, it does two things:
MergeSortExchangeExecutor2 takes inputs from multiple sources and
The outputs of all the sources have been sorted in the same way.
MySqlQuery executor. Runs a query against a MySql database.
Nested loop join executor.
PostgresQuery executor. Runs a query against a Postgres database.
Id of one row in chunked data.
Executor that scans data from row table
Range for batch scan.
SortAggExecutor implements the sort aggregate algorithm, which assumes
that the input chunks has already been sorted by group columns.
The aggregation will be applied to tuples within the same group.
And the output schema is [group columns, agg result].
Sort Executor
SortOverWindowExecutor accepts input chunks sorted by partition key and order key, and
outputs chunks with window function result columns.
Top-N Executor
UpdateExecutor implements table update with values from its child executor and given
expressions.