Infrastructure

What runs under your services.

Concrete details about the platform that hosts, scales and protects your apps. No hand-waving.

Compute

Container runtime and isolation

Runtime: Our own scheduler — no Kubernetes overhead, built specifically for containers billed by resources.
Container engine: OCI-compatible containers built from your Dockerfile, a compose file, an uploaded archive or an auto-detected stack.
Tenant isolation: Each service runs in its own container with kernel namespaces, cgroup limits and a private network per project.
Hardware: AMD Ryzen 9 up to 5.7 GHz, DDR5 memory, NVMe storage. x86_64 only.
Resource sizing: vCPU and RAM are set manually per service and can be changed at any time. There is no autoscaling today.

Network

Edge / CDN: Cloudflare in front of all public traffic.
TLS termination: Managed Let's Encrypt certificates, issued and renewed automatically for every domain, subdomain and wildcard.
Private networking: Your services reach each other over an internal network without going through the public internet; private services have no public address at all.
DDoS protection: L3/L4 and L7 filtering at the Cloudflare edge, enabled automatically on every plan at no extra cost.

Storage

Block storage: NVMe SSD persistent disks mounted directly into your container.
Managed databases: PostgreSQL, Redis and MongoDB on current stable major versions, with credentials and a connection string issued on creation.
Backups: Scheduled and on-demand backups for both volumes and databases, with a file list and restore from any saved copy.
Backup retention: Backups are kept for 7 days, including after a service or database is deleted.

Availability

Uptime target: We target 99.5% monthly uptime. There is no formal SLA with credits yet — we would rather state the target honestly than sell a contract we cannot back.
Service recovery: Containers are restarted automatically when they exit; you can also stop, restart, redeploy or rebuild any service from the dashboard.
RPO / RTO: RPO (how much data you can lose): up to 24 hours for databases on daily backups. RTO (how long recovery takes): up to 1 hour for a full restore.
Database HA: Managed databases run as a single node with automatic restart and backups. Replicas with automatic failover are not available today.

Operations

Logs: Build and runtime logs stream live in the dashboard. Retention is limited; there is no external log shipping yet.
Metrics: CPU and RAM usage per service with history.
Deploy history: Every deploy is recorded with status, duration and build logs. Redeploy of the current configuration is available; switching back to an older build is not.
Scheduled jobs: Cron schedules per service: create, edit, enable, disable or run once manually.

Security

Environment variables: Environment variables are managed in the dashboard and injected into the container at runtime. There is no separate secrets vault product.
Build isolation: Builds run in disposable containers; images are stored in a private registry scoped to your account.
Network access control: Private services are unreachable from the internet; databases accept connections only from your own services.
Shell access: No SSH or web shell into running containers — you work through deploys, env vars, logs and metrics.
Audit logs: Deploy and configuration history is available per service; full account audit logs are not implemented yet.

Compliance

Certifications: Muerte Cloud itself is not SOC 2 or ISO 27001 certified.
Subprocessors: Cloudflare for edge traffic and Selectel (ООО «Селектел») as the accredited hosting provider whose data centres run the platform. A DPA is available on request.
Hosting provider: The platform runs in the data centres of ООО «Селектел» (Selectel), a hosting provider listed in the Roskomnadzor hosting provider registry. Muerte Cloud is the software layer on top of that infrastructure.
What we don't claim: No formal SLA, no public status page, no multi-region deployment and no customer-facing object storage. When those ship, this page changes first.

Trust

Ask us directly — we'd rather tell you the honest limits of the platform than have you find them in production.