Epure benchmarks: footprint and RAM labs
Lab pages for compose footprint and dated RAM receipts. Not a second alternatives list.
On this pageShowHide
What belongs here
- Verifiable metrics with a method and a date.
- Architecture deltas that explain why numbers differ.
- Links into install, Verify, and compare pages.
Not here: undated vanity RAM headlines, 100% compatible claims, or SaaS mush.
Child pages
Sentry self-hosted RAM
- Job
- Official Sentry floor vs Epure dated idle receipts
Two-container
- Job
- Compose footprint scoreboard (2 vs 20+ / about 4 / 1)
Small VPS
- Job
- Small-box framing for error tracking self-host
| Dimension | Job |
|---|---|
| Sentry self-hosted RAM | Official Sentry floor vs Epure dated idle receipts |
| Two-container | Compose footprint scoreboard (2 vs 20+ / about 4 / 1) |
| Small VPS | Small-box framing for error tracking self-host |
Open the three labs below. Do not treat this table as a measurement.
Sentry self-hosted RAM. Two-container. Small VPS.
How to read receipts
- Different hosts are different rows. Do not average.
- Idle RSS is not a capacity guarantee under flood.
- Re-measure with
docker statson your metal after compose is healthy.
Dated Epure idle examples, also on the RAM child: 2026-09-23, 2 vCPU / 769 MiB Linux VPS (Alibaba Cloud), about 53 MiB combined (Epure about 5 + Postgres about 48). 2026-09-12, Docker Desktop, about 82 MiB app + Postgres combined.
Next steps
- Install or Self-host.
- Cheap VPS install.
- Point a DSN: Quickstart.
- Compare product shape: vs Sentry.
Questions
Why not put this under the blog?
Benchmarks are lab evidence. Essays stay on the blog. Avoid remaking RAM angles as new posts.
Can I quote 82 MiB everywhere?
Only with the dated Docker Desktop receipt (2026-09-12), or quote the 2026-09-23 VPS row (about 53 MiB). Otherwise say two-container or cheap VPS.
Is Cloud measured here?
No. Cloud is not live.