Postgres, highly available.
Public prices.
Production Postgres with automatic failover, backups and connection pooling — created and managed by API or your AI agent. The predictable alternative to AWS RDS, Neon and Railway: priced per reserved GB of RAM and disk, by the hour, shared vCPU included.
RDS is powerful — and a bill you can’t predict
AWS RDS charges for the instance, plus CPU credits when it’s busy, plus RDS Proxy for pooling, plus storage, backups and egress — a small instance can quietly cost more in CPU credits than in the instance itself. And you still wire up failover, pooling and monitoring yourself.
How koigrid compares
| koigrid | AWS RDS | Neon | Railway | |
|---|---|---|---|---|
| High availability (auto-failover) | Included | Multi-AZ (2×) | Paid tiers | Add-on |
| Connection pooling | Built-in (PgBouncer) | RDS Proxy (extra) | Built-in | Limited |
| Backups + point-in-time restore | Included | Included | Included | Included |
| Pricing | Reserved resources, per copy | Instance + credits + proxy + egress | Usage-based | Usage-based |
| Create/manage by API or AI agent | Native | Complex | API | API |
From public pricing, 2026. HA = leader + standby with automatic failover (Patroni + etcd).
What you’d actually pay
A production Postgres with HA + pooling + backups:
koigrid Postgres HA
$36.80/mo
2 copies × (1 GB, 0.5 vCPU, 20 GB) — pooling + backups included
AWS RDS
~$58/mo
instance + CPU credits + RDS Proxy + backups
A price you can compute in advance — no CPU-credit surprises, no per-connection proxy bill.
Based on a real RDS bill: on a busy small instance, the CPU credits alone can exceed the instance cost.
Production-grade by default
Automatic failover
Leader + standby on separate hosts with Patroni + etcd consensus. The primary dies → a standby is promoted in seconds, no data lost.
Connection pooling
PgBouncer in front of every database — no connection exhaustion from serverless or AI agents.
Read replicas
A separate read-only endpoint load-balances across the replicas — offload reads and scale without touching the leader.
Backups + PITR
Continuous WAL archiving to object storage. Restore to any point in time.
Automatic minor upgrades
PostgreSQL patch releases applied for you in a night maintenance window (UTC), on by default like RDS. Change the window or turn it off by API.
Branching
Clone your production database to test a migration or feature in isolation — like Neon branches. Reset a branch back to production (atomic swap), ephemeral TTL + delete protection, all by API.
Roles
Manage least-privilege Postgres roles per database (never superuser) for scoped, safe access — passwords never touch logs. Great for AI agents.
Agent-native
Create, back up and manage databases from Claude Code or any LLM via the API.