A high-performance distributed storage system built for the modern data era.
KDSS was created to address the growing gap between storage capacity demands and infrastructure budgets. Traditional 3-way replication wastes 200% of raw storage capacity. By leveraging Reed-Solomon erasure coding at its core, KDSS achieves up to 90.6% space efficiency while maintaining the same level of data durability -- enabling organizations to store more data with fewer disks, lower power consumption, and reduced operational costs.
Built with proven, high-performance technologies.
Core language for all services. High concurrency, low latency, single binary deployment.
High-performance RPC framework for inter-node communication with streaming support.
Efficient binary serialization for metadata and wire protocol definitions.
Flexible document store for metadata management with replica set support.
Erasure coding algorithm for data protection with configurable redundancy levels.
Embedded key-value store for local index and cache management on storage nodes.
Filesystem in Userspace for POSIX-compatible mounting without kernel modules.
Metrics collection and alerting with 32 built-in alert rules for cluster monitoring.
Pre-built dashboards for real-time visualization of cluster health and performance.
Upcoming features and milestones for the KDSS project.
Asynchronous cross-region replication for disaster recovery. Replicate EC-encoded data across geographically distributed clusters with configurable consistency policies.
Automatic data lifecycle management with hot/warm/cold tiers. Migrate data between NVMe, SSD, and HDD based on access patterns and configurable policies.
Native Kubernetes operator for declarative cluster management. Automated scaling, rolling upgrades, and CSI driver integration for persistent volumes.
Server-side encryption with AES-256-GCM. Support for external key management systems (KMS) including HashiCorp Vault and AWS KMS integration.
Join the KDSS community and help shape the future of distributed storage.