Overview
Architecture
Supercharge memory, I/O, communication, and virtualization services for SuperPoDs with a scale-up programming model
It provides resource pooling and dynamic scheduling for memory and DPU resources, supports distributed automatic primary node selection, and implements N-1 HA. It serves as the core control-plane reference implementation for the UB computing system.
It supports virtualization pooling, live migration policy decision-making, quick disaster recovery, and efficient VM/container communication, improving virtualization performance.
It supports unified memory semantics programming to implement memory sharing and pooling of UB-based SuperPoDs.
It provides high-performance, high-reliability, and ecosystem-compatible (user-space socket/Verbs over UB) communication protocols built on UB-based SuperPoDs.
It provides high-level I/O services for application-affinity global data read/write caching systems (such as SSU passthrough and KV cache) based on SuperPoDs.
