Add ability to estimate file size lazily #51
Merged
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Currently the
RecordBatchingSinker
checks whether a batch should be finalized after processing every record, to do that it passes the current record count and the current file size to a givenFileCommitStrategy
. The problem is that estimating the file size for formats likeparquet
can be very expensive, we observe ~15% of total time spent there. In addition, the file size in case ofparquet
is only an estimate, so doing it so frequently is pointless.We thus change the interface of
FileCommitStrategy
to accept the file size lazily, i.e. if the strategy does not invoke it, it will not be calculated. In addition we tuneFuzzyReachedAnyOf
to have a configurablefileSizeSamplingBatchSize
, i.e. a number of records processed that triggers a file size check (e.g. every 1000 records). By default it checks every time so that the default behavior remains the same.