# Shipyard: Building Slack's Next-Gen EC2 Platform

Managing EC2 instances with modern infrastructure principles

- **Source:** [Engineering at Slack](https://brifing.si/source/9869c8d3-df06-4af5-8add-0149a9b983ef)
- **Published:** 2026-07-14
- **Topic:** [Software & Engineering](https://brifing.si/c/engineering)
- **Original article:** <https://slack.engineering/shipyard-how-we-built-slacks-next-generation-ec2-platform/>
- **On Brifing:** https://brifing.si/s/e4500b20-c715-458c-9571-a92218409c59

## Introduction

Slack transitioned from a Chef-based legacy system to a more secure and scalable infrastructure for managing thousands of EC2 instances. The first step involved implementing multi-stack Chef deployments with version control. However, the limitations of continuously updating long-lived EC2 instances became apparent.

## Legacy System Challenges

Continuous updates made service deployments difficult and led to infrastructure drift. Coordinating changes across multiple layers created complexity. While containers offered solutions for some workloads, migrating the entire system wasn't practical.

## Birth of Shipyard

The new platform treats EC2 instances as deployable artifacts. It brings modern application deployment principles like immutable infrastructure, progressive rollouts, and automated safety checks to EC2. This approach provides teams with service-level deployment primitives.

## Multi-Architecture Support

Shipyard supports both AMD64 and ARM-based Graviton instances for cost and performance optimization. It works with multiple operating systems including Ubuntu, RHEL and Amazon Linux. Ideal for workloads like Kubernetes worker nodes that can't migrate to containers.

## Metrics-Driven Deployments

Integrated with Gondola orchestration system, Shipyard can automatically pause or rollback based on service health signals. Progressive deployments dramatically reduce failure impact radius.

## Fast Provisioning

Using a container-inspired layered image approach, shared 'golden' base images serve as foundations for service-specific images. This enables quick and predictable instance launches across regions.

## Simplified Configuration

Configuration management shifts from background processes to defined lifecycle phases like image baking. This makes system behavior easier to understand as instances aren't continuously changing.

## Real-Time Inventory

The new Peekaboo system provides real-time visibility into EC2 fleet state. Built with AWS EventBridge, OpenSearch and Lambda, it offers centralized monitoring of all instances.

## Short-Lived Instances

For security and immutability, each EC2 instance has a limited lifespan and gets automatically rotated. This keeps fleets fresh and limits potential vulnerability impacts.

## Slack-zero Base Image

At Shipyard's core lies the 'slack-zero' shared base image. Created by Slack's Compute Platform Team and maintained with security teams, it contains the core OS and common infrastructure components.

## Team Impact

The new platform fundamentally changes how teams update infrastructure—deploying new images instead of modifying instances. This paradigm shift reduces operational overhead while increasing deployment safety.

## Conclusion

Shipyard represents a fundamental shift in how Slack runs EC2 instances. By bringing modern deployment practices to EC2, it maintains flexibility while enhancing security and scalability. This transformation sets a new standard in cloud infrastructure management.

## Sources covering this story

- https://slack.engineering/shipyard-how-we-built-slacks-next-generation-ec2-platform/ — 

---

_Summary written by Brifing from the original reporting. Copyright in the original article belongs to its publisher; read the full piece at the link above._