A practical guide to Postgres-backed jobs

Same database.
Different promises.

Choose a queue by the work it must protect. Explore job frameworks, durable workflows, visibility queues and event queues—with the evidence kept in view.

14systems in this guide
8benchmark adapters
4different system contracts
No universal ranking.Documentation, measured results and simulations stay separate.

Benchmark runs · 2026-11-11 to 2026-11-15

Measured under the same load.

All results, method and raw data →

9 adapters, one host, PostgreSQL 18.6. Each scenario answers one question; there is no overall score.

Throughput at saturation

How many 1 ms jobs per second does one worker process complete as it gets more workers?

Tuned: River 10.9k and Awa 10.8k jobs/s; PgQue 47k events/s, producer-bound.

Jobs/s, 64 workers · best 3 of 8
  1. PgQueevent batches18.9K jobs/s
  2. Awa5,689 jobs/s
  3. PGMQvisibility queue5,379 jobs/s
See the results →

Latency at a steady 800 jobs/s

At a rate every system should sustain, how quickly does a job go from insert to completion?

Awa, Oban and PGMQ pick up within 20 ms p99; three systems fall behind.

Pickup p99 at 800 jobs/s · best 3 of 8
  1. PGMQvisibility queue7 ms
  2. Awa19 ms
  3. Oban20 ms
See the results →

Cost of doing nothing

With workers running and no jobs, how much does each system cost the database?

16–24 xacts/s for most; PGMQ polls at 160/s; Awa holds 276 MB of Postgres memory.

Idle transactions/s · best 3 of 8
  1. Oban16 xacts/s
  2. pg-boss21.9 xacts/s
  3. Absurdworkflow23.5 xacts/s
See the results →

PostgreSQL restart under load

When the database restarts mid-run, does the system come back and work off what accumulated?

All eight recover; Absurd and Procrastinate need a supervisor to restart the worker.

Jobs/s after a Postgres restart · best 3 of 8
  1. PgQueevent batches805 jobs/s
  2. Awa802 jobs/s
  3. Absurdworkflow802 jobs/s
See the results →

Impact on a neighbouring application

Does a busy queue slow down unrelated transactions on the same server?

Neighbour throughput unchanged everywhere; p99 at 4× from 2.4 to 18 ms.

Neighbour p99, queue at 4× · best 3 of 8
  1. PgQueevent batches2 ms
  2. Procrastinate5 ms
  3. Awa6 ms
See the results →

Logical replication alongside the queue

Does the queue coexist with CDC on the same database, including when the CDC consumer stops?

Awa 19 → 0 tables without replica identity; still the most catalog churn.

Dead catalog tuples, stalled CDC · best 3 of 8
  1. PGMQvisibility queue20 tuples
  2. Absurdworkflow22 tuples
  3. Procrastinate25 tuples
See the results →

A herd of scheduled jobs

When 20,000 jobs come due at the same instant, how late do they start, and do immediate jobs suffer?

PGMQ 1.9 s, Awa 12 s, Oban 21 s to start a 20,000-job herd.

Scheduled herd lateness p99 · best 3 of 7
  1. PGMQvisibility queue1.94 s
  2. Awa12 s
  3. Oban20.5 s
See the results →

A storm of failing jobs

When a third of jobs start failing, does useful work keep flowing?

Every system kept the good jobs flowing through a failure storm.

Good jobs/s in a failure storm · best 3 of 8
  1. Absurdworkflow139 jobs/s
  2. PgQueevent batches138 jobs/s
  3. PGMQvisibility queue138 jobs/s
See the results →

Many queues

What does spreading the same load over 200 queues cost, and how does each system serve many queues?

All keep up; idle cost at 200 queues from 0.02 to 1.7 Postgres cores.

Pickup p99 over 200 queues · best 3 of 8
  1. PGMQvisibility queue3 ms
  2. pg-boss5 ms
  3. Procrastinate11 ms
See the results →

Large job arguments

What happens with 64 KiB and 1 MiB arguments, and does the database shrink back afterwards?

Awa shrinks back to 21 MB after 8 GiB; others keep 7–16 GB.

Database after an 8 GiB drain · best 3 of 8
  1. Awa21.3 MB
  2. Procrastinate7.35 GB
  3. River8 GB
See the results →

Moderate payloads over time

At a gentle rate with 64 KiB arguments, how much WAL is written and how much space is left behind?

Only Awa returns the space (22 MB vs 1.2 GB).

See the results →

Long-running jobs

What does it cost to keep jobs running for 40 seconds each?

WAL to keep 40 s jobs alive: 4.6 KB/s (PGMQ) to 28 KB/s (Awa).

WAL/s with 40 s jobs · best 3 of 8
  1. PGMQvisibility queue4.64 KB/s
  2. Procrastinate5.95 KB/s
  3. PgQueevent batches6.26 KB/s
See the results →

Draining a backlog

How fast does a 200,000-job backlog drain, and how big is the database afterwards?

PGMQ and PgQue drain at ~14.5k jobs/s; only Awa shrinks back.

Backlog drain rate · best 3 of 7
  1. PGMQvisibility queue14.5K jobs/s
  2. PgQueevent batches14.3K jobs/s
  3. Awa5,154 jobs/s
See the results →

Finding one job among a million

With a million jobs queued, how long does an operator wait to find one?

River finds one job in 2 ms; Awa 0.6 times out; Procrastinate stalls behind the haystack.

See the results →

Pausing and resuming a queue

When an operator pauses a busy queue, does work stop everywhere, and how fast does it come back?

A pause survives a restart only where it's stored: Awa, River, PgQue.

See the results →

A million abandoned claims

When an entire fleet dies holding about a million claimed jobs, how long until those jobs have run?

PgQue 16 s, PGMQ 5.9 min, pg-boss 14 min, Awa 20.5 min to finish 1M abandoned jobs.

1M abandoned jobs, all done again · best 3 of 6
  1. PgQueevent batches16.3 s
  2. PGMQvisibility queue5.88 min
  3. pg-boss14 min
See the results →

The directory

Start with your constraints.

What’s covered, and what isn’t ↗

14 systems · alphabetical order

Choose up to 8 to compare
Durable workflow

Absurd

TypeScript · Python · Go (experimental)

Checkpointed workflows that replay persisted steps after interruption.

Adapter in this repoMeasured 0.5.1
Read profile
Job framework

Awa

Rust · Python

Typed Rust and Python jobs with segmented storage and guarded attempts.

Adapter in this repoMeasured 0.6.17+git.1a53fd9eDocs: development snapshot
Read profile
Job framework

A Rails ActiveJob backend with a dashboard and concurrency extensions.

Documentation only
Read profile
Job framework

Graphile Worker

Node.js · TypeScript

Node.js task executors, SQL enqueue and serial named queues.

Documentation only
Read profile
Job framework

Oban

Elixir

An Ecto and OTP job framework with queues, retries and telemetry.

Adapter in this repoMeasured 2.24.2
Read profile
Job framework

pg-boss

Node.js · TypeScript

Node.js job queues with explicit queue policies and transaction adapters.

Adapter in this repoMeasured 12.38.2
Read profile
Visibility queue

PGMQ

SQL · Language-neutral clients

SQL messages with a visibility timeout and application-owned consumers.

Adapter in this repoMeasured v1.13.1
Read profile
Event queue

PgQ

SQL · C · Python ecosystem

The maintained original PgQ event engine, with tick-based batches.

Documentation only
Read profile
Event queue

PgQue

SQL · Go · Python · TypeScript

A shared event log with consumer cursors and batch acknowledgements.

Adapter in this repoMeasured v0.3.0-alpha.1-30-g8e3064cDocs: early-stage api
Read profile
Job framework

PotatoQ

Python

An alpha Python worker with Celery-shaped APIs and a Postgres transport.

Documentation onlyDocs: alpha
Read profile
Job framework

Python sync/async tasks with application locks and configurable recovery.

Adapter in this repoMeasured 3.11.0
Read profile
Job framework

Que

Ruby

A focused Ruby queue built around advisory locks and transaction control.

Documentation only
Read profile
Job framework

River

Go · Ruby · Rust · TypeScript

A PostgreSQL job framework with a shared format across language clients.

Adapter in this repoMeasured v0.51.2
Read profile
Job framework

Rails job infrastructure with SQL dispatch and explicit failure records.

Documentation only
Read profile

Language labels describe the inspected API/SDK surface, not equal maturity or feature parity across bindings. Read each profile’s scope and snapshot.