mirror of
https://github.com/lancedb/lancedb.git
synced 2026-11-14 06:36:54 +00:00
Jack YeandClaude Sonnet 5
de95ed3f6d
A broad documentation pass across the areas LanceDB Enterprise and Embedded diverge on most: **Data Curation** -- GPU Function execution, Slurm scheduling (and what Slurm is, for readers new to it), blob late materialization as its own page, joins, aggregation/profiling, sampling, near-duplicate detection, chunking, distributed shuffle, and IVF-partition clustering. **Feature Engineering** -- computed columns and materialized views, restructured to lead with the Function-based path and show the SQL form side by side with Python throughout; a single Auto Refresh page instead of duplicating scheduling details on every refresh page; adaptive execution; checkpointing; GPU and Slurm placement for refresh workers; a Training Cache page for the PyTorch data loader's automatic cache benefit. **Security & Governance** -- an Audit Log reference (schema, access, grants), Secrets promoted to an ordinary catalog object with their own SQL DDL and a vending/rotation guide, and privilege-table fixes verified against the authorization source. **Deployment & Maintenance** -- Horizontal Scaling and OpenTelemetry moved into a real Deployment area; Automated Cleanup folded into Intelligent Auto-Maintenance (every other maintenance decision already lived there); RTE Compaction and Advanced Conflict Resolution documented in depth. **Landing page** -- the Enterprise feature comparison table reorganized (several redundant or stale rows removed, several new areas added), and the "why Enterprise" bullets replaced with ones tied to a concrete page (Performance, Scale, Fault tolerance, Multicloud deployment, Simplify operations) instead of vague promises. Plus a full pass addressing lancedb-gatekeeper's review: a real test-safety bug (a quickstart test could silently overwrite and delete a deployment's own `frames` table), a fabricated SQL statement with no backing implementation, an actual dedup-guarantee bug in `take_row_ids`, and dozens of documentation-accuracy fixes verified against source rather than taken on faith -- each either fixed, or left with an explicit note on why it needs a feature or deployment owner's sign-off instead of being guessed at. 🤖 Generated with [Claude Code](https://claude.com/claude-code) https://claude.ai/code/session_018PXMCbRH6i5w47TsjeEUMT --------- Co-authored-by: Claude Sonnet 5 <noreply@anthropic.com>
64 lines
2.3 KiB
Plaintext
64 lines
2.3 KiB
Plaintext
---
|
|
title: "Distributed Shuffle"
|
|
sidebarTitle: "Distributed Shuffle"
|
|
description: "How a SQL query's intermediate results move between plan executors for a join, aggregate or sort."
|
|
icon: shuffle
|
|
availability:
|
|
embedded: unavailable
|
|
enterprise: available
|
|
---
|
|
|
|
A distributed SQL query plans as a sequence of **stages**, each with its
|
|
own set of partitions, connected by an **exchange**: a streaming gRPC
|
|
service that carries one stage's output to the next stage's plan
|
|
executors. [Distributed search](/enterprise/distributed-search) covers
|
|
how a query gets placed on the fleet in the first place; this page covers
|
|
what happens between placement and the final result, for a query whose
|
|
plan needs data repartitioned across machines -- a hash join or a grouped
|
|
aggregate, for example, where rows with the same key must land on the
|
|
same worker regardless of which plan executor originally read them.
|
|
|
|
```mermaid
|
|
flowchart LR
|
|
subgraph Stage1["Stage: scan"]
|
|
direction TB
|
|
PE1["Plan executor<br/>reads its partition"]
|
|
PE2["Plan executor<br/>reads its partition"]
|
|
end
|
|
|
|
EX["Exchange<br/>streaming gRPC,<br/>repartitions by join/group key,<br/>backpressure via a byte budget"]
|
|
|
|
subgraph Stage2["Stage: join / aggregate"]
|
|
direction TB
|
|
PE3["Plan executor<br/>probes its share of keys"]
|
|
PE4["Plan executor<br/>probes its share of keys"]
|
|
end
|
|
|
|
PE1 --> EX
|
|
PE2 --> EX
|
|
EX --> PE3
|
|
EX --> PE4
|
|
```
|
|
|
|
## Pipelined, not batched {#pipelined}
|
|
|
|
The exchange moves data as a gRPC stream of row batches, not a file a
|
|
later stage waits for: a consuming plan executor pulls batches as they
|
|
become available, and a slow consumer throttles its producer through a
|
|
byte budget on unread messages rather than letting memory grow without
|
|
bound. A partition that stops being read -- for example the probe side of
|
|
a hash join whose build side turned out empty -- is drained rather than
|
|
left to stall the budget for every other partition sharing the same
|
|
connection.
|
|
|
|
## What to read next {#next}
|
|
|
|
<CardGroup cols={2}>
|
|
<Card title="Distributed search" icon="network-wired" href="/enterprise/distributed-search">
|
|
How a query gets placed on the query-serving fleet in the first place.
|
|
</Card>
|
|
<Card title="Query plans" icon="map" href="/basics/sql#query-plans">
|
|
Inspect a specific query's plan and its actual execution metrics.
|
|
</Card>
|
|
</CardGroup>
|