Skip to main content

What is the Coordinator?

The Coordinator is the control plane of the DS3 Composer platform. It orchestrates metadata-related operations, including shard placement, system health monitoring, authentication, and recovery workflows. The Coordinator exposes a set of APIs and a graphical dashboard, enabling operators to configure and monitor distributed storage environments at scale. While the DS3 architecture is fully distributed, the Coordinator provides a lightweight, centralized layer for managing critical orchestration tasks across a Swarm. It does not store user data and is compatible with zero-knowledge encryption modes. By centralizing this function within the Coordinator, DS3 Composer ensures predictable growth, operational scalability, and consistent enforcement of platform-wide policies.

Why the Coordinator Matters​

  • Metadata Management: Coordinates object lifecycle metadata such as shard placement, health status, and policy enforcement.
  • Resilient by Design: Can be deployed across multiple sites or clouds to maximize uptime and eliminate single points of failure.
  • Zero-Knowledge Capable: Fully compatible with DS3's privacy-preserving encryption model.
  • Multi-Tenant Friendly: Designed to orchestrate multiple Swarm installations simultaneously, making it ideal for service providers and large-scale enterprise setups.

Deployment Options​

The Coordinator can be deployed in two main modes:

1. Managed (SaaS) by Cubbit​

  • Default option for most users
  • High availability and fault-tolerance
  • Operated by Cubbit with predefined SLAs
  • Ideal for users who prefer hands-off orchestration

2. Self-Hosted (On-Premises)​

  • Recommended for air-gapped environments
  • Requires a license
  • A full installation guide is currently under preparation and is not yet available publicly
  • Supports:
    • Active-Passive clustering for high availability (HA)
    • Master-slave architecture across one or two sites
    • Stretch clustering to a third site for advanced business continuity
    • Backup via DB dumpfor disaster recovery (asynchronous, non-zero RPO)

All deployment scenarios are engineered to avoid reliance on proprietary third-party tools, enabling portability and simplifying long-term maintenance.

Shared Coordinator Model​

The Coordinator is designed to support multi-tenant orchestration, meaning:

  • A single instance can serve multiple Swarm deployments and multiple Tenants
  • Operational costs (e.g., replication, backup, monitoring) can be amortized across installations

This model enables highly cost-effective coordination in multi-cluster or service-provider scenarios, requiring centralized governance over many isolated storage environments.

Last updated on