Balta Engineering
How we recover a service when a host fails
Reserved capacity, verified backups and a connection address that doesn't change.
- reliability
- architecture
- sla
How it works, in detail.
Technical writing about the parts of a managed database that are usually described in adjectives. Every claim here is checkable against our documentation.
- Balta Engineering
Why we keep a host's worth of capacity free in every location
The reserve capacity behind every Balta recovery.
- capacity
- reliability
- operations
- Balta Engineering
Pooled or direct: choosing the right Postgres endpoint
When to use connection pooling, and the session features that need a direct connection.
- connections
- pgbouncer
- postgres
- Balta Engineering
A backup you've never restored is a hypothesis
Why we restore every customer's backups on a schedule — and what it tells you.
- backups
- recovery
- verification
- Balta Engineering
What "4 GB of RAM" actually means on Balta
How we enforce memory in the kernel, and why it matters for performance.
- memory
- cgroups
- postgres