Octohost

Octohost — Game hosting without the nonsense.

Overview

Octohost was a managed multiplayer game server hosting company I co-founded in 2017. The core insight: spinning up a fast, reliable modded Minecraft server required a VPS, Linux administration, JVM tuning, and ongoing ops. Most incumbents “solved” this with heavily oversubscribed shared hosting — cheap, but unreliable and slow under load.

Octohost took a cloud-engineering approach: provide dedicated, performance-oriented servers with automation and sane defaults so customers could go from purchase → playable server without learning infrastructure.

What I built (cloud engineering)

Automated provisioning & deployment

• Built the deployment automation and provisioning flow so a paid order could trigger repeatable server creation , configuration, and handoff to the customer.
• Standardized server setup to reduce “works on this box” drift and make rollouts safer and faster.

Containerized infrastructure

• Designed and operated a distributed game server platform using Docker and Kubernetes , focused on:
  • Low-latency gameplay
  • Isolation between customers
  • Predictable resource allocation (CPU/RAM/IO) per server
  • Faster iteration on platform changes without interrupting the customer experience

Reliability & performance as product features

• Established operational patterns to keep performance stable under noisy-neighbor pressure:

  • Resource limits / scheduling strategies
  • Deployment discipline (canary-style changes where possible)
  • Clear incident-response loops and on-call-style debugging when production went sideways

• Tuned the end-to-end path from checkout → provisioned server to minimize time-to-play.

Billing model innovation (weekly plans)

Most of the industry assumed monthly billing. Octohost pioneered weekly billing , which meant the platform had to support:

• Shorter-lived instances
• Higher churn / more frequent provisioning events
• Cleaner lifecycle management (create, resize, delete) without leaving customers stranded

Architecture (high level)

• Control plane : Order + provisioning workflow that turns an order into infrastructure changes.
• Compute plane : Kubernetes cluster(s) running isolated workloads for customer servers.
• Runtime : Container images + configuration, with deployment automation to roll out updates safely.
• Ops : Monitoring/debugging practices to quickly identify hot spots (CPU, memory pressure, disk contention, network saturation) and recover without broad downtime.