Skip to main content
Version: Next

v2.6

This document outlines the roadmap for Dragonfly v2.6, focusing on performance optimization, enhanced functionality, and expanded use cases in AI/ML workloads. Dragonfly v2.6 is scheduled for release on December 31, 2026.

Core Components​

Manager​

  • Enhance service performance and resource utilization while reducing CPU/Memory overhead.
  • Enhance user experience and UI design in the Manager Console.

Scheduler​

  • Enhance service performance and resource utilization while reducing CPU/Memory overhead.
  • Optimize the scheduling algorithm to improve bandwidth utilization in the P2P network.

Client​

  • Enhance service performance and resource utilization while reducing CPU/Memory overhead.
  • Dfcache/Dfstore support to import persistent cache task to Dfdaemon in Node by UDS.
  • Support lightweight Client with no need to deploy centralized Scheduler and Manager.

AI Model/Dataset Distribution​

  • Writes use Direct IO, reads use Buffered IO.
  • Implement Python SDK to provide data distribution for AI Infrastructure.
  • Explore integrating the SGLang and vLLM ecosystems to distribute data.

AI Agent​

  • Ensure data reliability when asynchronously writing to object storage.
  • Implement Python SDK to support snapshotter for use in AI agent.
  • Explore integrating the Agent Sandbox, Gymnasium, etc.
  • Implement model context protocol (MCP).

Others​

Observability​

  • Improve and refine the monitoring metrics system.
  • Optimize the alerting mechanism and enhance issue diagnosis capabilities.

Security​

  • Implement encrypted data storage.

Testing​

  • Add more E2E tests and unit tests.

Documentation​

  • Enhance the landing page UI.
  • Add more documentation on system interactions and implementation details.

Nydus​

Testing​

  • Increase unit test coverage target to 60%. Consider leveraging agent capabilities.

Core Components​

Nydusd​

  • Deprecate erofs+fscache solution and migrate to erofs+fanotify pre-hook solution.

Snapshotter​

  • Further enhance observability. For example:
    • Collect statistics on nydusd image-related information.
    • Support Prometheus metrics collection.
  • Regarding Containerd's issues related to multi-snapshotter switching, organize best practice documentation.
    • follow Containerd community progress

Nydus Image​

  • Better support for image conversion from Nydus to OCI. Fix errors during reverse conversion of large images.

Kata Container Support​

  • Best practice documentation for using nydus in Kata Container scenarios.