Curvine 2025 Roadmap

Overview

This roadmap outlines the goals our team is pursuing and our vision for the entire community. The evolution of Curvine in 2025 is marked by a series of major feature updates. These enhancements stem from community feedback and the latest demands in real-world scenarios, reflecting our ongoing commitment to excellence.

Join Us

• For existing tasks with trackable issues, you can directly comment on them in GitHub Issues
• For open suggestions, please submit them through GitHub Discussions
• We are committed to developing an open-source ecosystem, and we welcome global contributors to join!

---

Core Feature Planning

1. Cache function improvement

• Cache data automatic cleanup
• Support mainstream storage systems：
  • HDFS
  • S3
  • OSS (OSS-HDFS)
• Cache data preloading

2. Data write consistency

• Write data automatically manages cache and refreshes to storage systems
• Pipeline multi-replica copy and maintenance
• Strong consistency protocol support

3. Shuffle support

• File aggregation function
• Integration with Spark RSS (Remote Shuffle Service)

4. Performance optimization

• ORPC uses io_uring to implement full-link zero-copy (solving the splice failure problem in cloud environments)
• FUSE full-link zero-copy (performance alignment with Client mode)
• RDMA network support

5. Cloud-native support

• Curvine Operator
• Curvine CSI driver

---

Version release plan

Milestones for 2025

Version number	Release date	Core features
0.1.1-beta	2025-07	Use Curvine FUSE as Spark Shuffle local disk (through TPC-DS test)
0.2.1-beta	2025-08	Shuffle function completion + Spark integration test
		io_uring acceleration for read/write
		HDFS protocol support
0.2.2-beta	2025-08	Cloud-native CSI driver support
0.3.1-beta	2025-09	Cache data automatic cleanup + multi-storage system adaptation

Major version planning

Version number	Time node	Target scene
1.0.0-base	2025-12-30	Comprehensive support for big data scenarios:
		- Shuffle optimization
		- Local Spill support
		- Hot data cache acceleration
2.0.0-base	2026-09-30	AI scenario enhancement:
		- Training acceleration framework integration
		- RDMA/GDS support