PLATFORM

Architecture built to hold under real traffic

Every Mirahost site runs as an immutable container on a multi-zone grid — isolated resources, health-checked failover, and an audit trail for every change.

The anatomy of a Mirahost site

A deploy — whether from Git or SFTP — becomes an immutable image: your runtime, your files, your configuration. The image runs as a container with reserved CPU and memory, attached to a managed database and fronted by the edge network.

01Edge layer — TLS termination, CDN cache, WAF rules
02Runtime container — PHP-FPM, Node, or Python with reserved resources
03Managed data — databases in separate containers, backed up independently
Edge layer TLS · CDN · WAF · 80+ locations Runtime container immutable image · reserved 2 vCPU / 2 GB health check every 10s · auto-restart Managed data layer MySQL / Postgres · nightly + on-deploy snapshots Each layer fails over independently — no single point of failure

What "managed" actually covers

RELIABILITY

Self-healing by policy, not heroics

Health checks every 10 seconds. A failing process restarts in place; a failing host reschedules the container to another zone. Three consecutive failed deploys trigger automatic rollback to the last healthy image — before your monitoring would have paged you.

SECURITY

Patched while you sleep

OS and runtime CVEs are patched at the platform layer within 24 hours of disclosure — your image is rebuilt on the fixed base and rolled out with zero downtime. Container isolation means a compromised neighbor is an architectural impossibility, not a firewall rule.

PERFORMANCE

Reserved, not "burstable"

CPU and memory are allocated per site from dedicated capacity. Traffic spikes scale your container vertically first, then horizontally behind the load balancer — within limits you set, at prices you already know.

COMPLIANCE

Audit-ready from day one

SOC 2 Type II reports on request, EU and US data residency options, immutable audit logs on every account action, and role-based access for your whole team. Compliance questionnaires stop being a week of work.

The numbers we hold ourselves to

View live status page →
99.98%
measured uptime, trailing 12 months
<10s
failure detection to recovery start
3
availability zones per region
30 days
backup retention, off-cluster

Uptime is measured externally by a third party and published incident-by-incident on our status page — including the bad days.

See it hold up under your workload

14-day free trial with production-grade resources — no card required.

Start free trial Read the docs