Append arrow directly to local data #100
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Currently any event that arrives is directly written to data.parquet. It is easier to do it this way but this ultimatly slows down log ingestion. To make log ingestion faster a stream of record-batches would act as an ideal buffer. Appending arrow record-batches directly to a stream is faster way to go about it but it has it's own challenges. - querying realtime would require having these record-batches in memory. - directly appending record-batches from multiple threads to an append only file is not supported by Arrow StreamWriter. Locking is used for now - Re-implementation of stats is required - Cannot fully guarantee state of localstorage if it failed exactly at move step of localsync. orphan tmp files can occur although that'd very rare
trueleo
force-pushed
the
stream_writers
branch
from
d395614 to
1af17bc
Compare
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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Outdated suggestions cannot be applied.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.
Description
Currently any event that arrives is directly written to data.parquet. It is easier to do it this way but this ultimatly slows down log ingestion. To make log ingestion faster a stream of record-batches would act as an ideal buffer. Appending arrow record-batches directly to a stream is faster way to go about it but it has it's own challenges.
also fixes #92
This PR has: