Add RFC for native ADBC connector in C++ workers - #64
Draft
jja725 wants to merge 1 commit into
Draft
Conversation
jja725
force-pushed
the
adbc-native-connector-rfc
branch
from
July 24, 2026 08:02
4da9cc1 to
9c90438
Compare
jja725
force-pushed
the
adbc-native-connector-rfc
branch
from
July 24, 2026 08:14
9c90438 to
45a3c6d
Compare
jja725
marked this pull request as draft
July 24, 2026 08:17
This file contains hidden or 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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Proposes a native ADBC connector for C++ workers: scan relational databases (MySQL, PostgreSQL, and other systems with ADBC drivers) directly in the worker process via Arrow ADBC, building on the generic ADBC connector added to Velox in facebookincubator/velox#18258.
The design is a point-for-point mapping of the existing
presto-base-jdbcflow: the coordinator reusesJdbcClientmetadata, dialect rules,TupleDomainpushdown, andJdbcComputePushdownunchanged, and renders the pushed-down predicate (with?placeholders and bound constants, exactly asQueryBuilderdoes today) at split-generation time. The C++ worker composes the final SQL from the split's predicate and the scan's column handles, executes it through an in-process ADBC driver, and imports the Arrow result stream into Velox vectors zero-copy. Becausebase-jdbclayouts report the constraint as unenforced, any predicate the ADBC path does not push stays correct — the engine re-applies it — which allows a safely phased rollout (inlined-literal subset first,AdbcStatementBindparameter binding for full parity later).This is a complement to RFC-0018 (Java connector federation), which addresses the same problem class through a Java Flight server. The RFC includes a detailed tradeoff comparison: federation can forward splits from any Java connector (RFC-0018 scopes its goals to single-node systems, but the split-forwarding mechanism is general) at the cost of a new service, a JVM in the data path, and a second network hop; the ADBC path covers the subset of systems with ADBC drivers and removes all three, at the cost of distributing credentials to workers and direct worker→database connection fan-out. The recommended posture is to adopt both, choosing per catalog.
Numbered RFC-0026 since 0024 and 0025 are claimed by open PRs (#59, #62, #63, #61).