Standard PostgreSQL on dedicated NVMe — isolated compute, not a shared pool. Backups,
PITR, connection pooling and monitoring are on from the first minute, tuned by someone
who does this for a living. Migrate in and out with pg_dump.
From $17/mo.
1. pick region, size, version
2. pay
3. postgresql://user@app.db…
Provision from the CLI, or from code — ephemeral per-agent databases, created and torn
down programmatically:
$ curl -X POST https://api.pgrun.dev/v1/databases \
-d '{"region":"fra","size":"small","version":"18"}'
#=> postgresql://... in ~120s
[02]
Pricing
Pick a size and go. Right-hand column is the same spec on RDS, for comparison.
hobby-1
standard-4
standard-16
on RDS
vCPU / RAM
1 / 2 GB
4 / 8 GB
8 / 24 GB
same spec
NVMe storage
16 GB
64 GB
64 GB
network EBS
backups + PITR
included
included
included
extra
pooling, replicas
included
included
included
extra
expert support
included
included
included
paid tier
price
$17/mo
$96/mo
$186/mo
—
Price is per month for the size plus its base storage; add storage per GB at checkout.
Cancel any time. Your data leaves with pg_dump — no export fee, no lock-in.
[03]
The platform
Fully-managed PostgreSQL, run by people who tune databases for a living. Everything
below is included — nothing to babysit.
+
Automatic backups
Off-site, daily, and tested so they actually restore.
+
Point-in-time recovery
Restore to any second — not just last night's dump.
+
Monitoring & alerts
Slow-query insights that catch problems before your users do.
+
One-click scaling
Online version upgrades, no downtime drama.
+
Connection pooling
PgBouncer built in — no separate service to run.
+
Read replicas
Spread reads and serve data closer to your users.
+
Encryption by default
At rest and in transit (TLS) — no extra config.
+
Free migration
From Heroku, RDS, Aurora, or Supabase — we do the moving.
+
40+ extensions
PostGIS, pgvector, TimescaleDB and the rest, without maintaining an image.
[04]
Latency
See what dedicated Postgres can change.
// p99 query latency — before and after migration
illustrative, from a 240GB workload we migrated in March — your numbers depend on
query design, data size and instance config
Fast storage built for high IOPS and low latency. No separate IOPS packages to buy.
+
Isolated compute
CPU and memory reserved for your database, protected from noisy neighbours.
+
Tuned for your workload
Configured around instance size, memory, connection limits, and how you actually query.
[05]
Questions
?How quickly can I launch a database?
Pick region, size and version, pay, and you get a connection string. We provision the instance, configure secure connections, and enable off-site backups. Most databases accept connections in about two minutes.
?Is there a trial or a refund?
Seven-day money back, no questions. Cancel from the dashboard and the charge is reversed.
?Is this standard PostgreSQL?
Yes — no proprietary layer, no application-specific API. Your database stays portable via pg_dump, pg_restore, and logical replication.
?What happens if pgrun shuts down?
You keep standard Postgres and daily off-site dumps you can download at any time. There is no format to convert and no export fee — pg_dump and point it at any other provider.
?Where are the servers?
Falkenstein and Helsinki (EU, GDPR jurisdiction) and Ashburn, Virginia (US). You choose the region at creation; data stays in it.
?Who has access to my data?
Only me, and only for an incident or a support request you opened. Access is logged, disks are encrypted at rest, and connections require TLS.
?Do you have an API or CLI?
Yes. Create, resize, snapshot and destroy databases over the REST API or the pgrun CLI — suitable for ephemeral per-agent databases in CI or agent runs.
?Which versions and extensions?
Supported production majors at creation time; new majors are added after upgrade safety is tested. More than 40 common extensions, depending on version and plan.
?How do backups work?
All databases include encrypted off-site backups. Eligible plans add point-in-time recovery within the configured retention window.
?Can you migrate my existing database?
Heroku, RDS, Aurora, Supabase, DigitalOcean, self-hosted. We review your extensions, traffic, and acceptable downtime before planning the cutover. Standard migrations are included.
?Will my application experience downtime?
Most migrations need only a brief cutover window. The exact length depends on size, write volume, and the replication method your current provider allows.
?Can you help with performance problems?
Yes — slow queries, indexes, connection limits, autovacuum, memory configuration, workload planning, and production incidents.
