Against the backdrop of deep integration between the digital economy and artificial intelligence, enterprises are increasingly demanding more agile, resilient, and intelligent integrated cloud infrastructure to underpin business innovation. As a leading domestic provider of cloud infrastructure solutions, KAYTUS officially announces the launch of the KSRail Hyper‑Converged Infrastructure System V8.
Built around the core design philosophies of AI Scenario-Optimized, Extreme Performance, and Full-Stack High Availability, KSRail V8 delivers breakthrough innovations for mission-critical scenarios including artificial intelligence, high concurrency, and multi-chip collaboration, empowering enterprises to build a future-ready intelligent cloud foundation.
Multiple Memory Overcommitment
KSRail V8 introduces an all-new memory overcommitment feature. By pooling and deeply interconnecting DRAM memory through memory pooling technology, and leveraging advanced cross-layer memory algorithm collaboration, it constructs a unified memory resource pool that delivers multiple-fold physical memory overcommitment capability. This significantly increases VM density per node while lowering total cost of ownership (TCO).
Smooth I/O Stack Fault Resilience
In the latest distributed storage (dSAN) architecture, KSRail V8 integrates a combined solution of Fault Pre-Diagnosis & Real-Time Perception, Data Replica Degraded Writing, and Dynamic Intelligent Synchronization. With full data safety guaranteed, it enables uninterrupted I/O during replica failures with performance degradation limited to <20%, elegantly resolving replica and slow-disk failure challenges.
Global Consistent PTP Clock Source
KSRail V8 pioneers global consistent PTP clock source technology. It supports hardware-level agentless PTP clock synchronization, innovative NIC driver optimization, visualized monitoring, and full compatibility with domestic PTP architectures, providing robust support for mission-critical workloads on the cloud.
2 Million IOPS
Through full-stack technological innovation and deep optimization, KSRail V8 achieves over 2 million IOPS in a 3-node hyper-converged cluster. It supports sub-second failover for single RDMA NIC failures, redefining the performance frontier for enterprise-grade data processing.
Ultra-Low I/O Latency
KSRail V8 delivers up to 600,000 IOPS per VM with stable disk latency of 0.2–0.4 ms, reliably running mission-critical applications such as databases, caches, message queues, and search engines.
vhost-Kernel I/O Acceleration
KSRail V8 leverages full-stack kernel-level I/O acceleration to boost VM performance by multiple times, with single-VM performance reaching 600,000 IOPS, continuously pushing performance boundaries.
Distributed Stretched Cluster
The KSRail V8 Hyper-Converged Stretch Cluster Solution adopts an innovative architecture and technological breakthroughs to create a new dual-site disaster recovery model featuring uninterrupted business continuity and fully automated disaster recovery. It supports combined disaster recovery modes, including hyper-converged + active-active centralized storage and SRM + centralized storage. A hierarchical disaster recovery system is built to achieve zero downtime for mission-critical services via stretch clusters and cost-effective cross-site takeover for non-mission-critical services via SRM, satisfying users’ disaster recovery requirements at all levels.
vTPM 2.0 Secure & Trusted Cloud
KSRail V8 adds virtual Trusted Platform Module (vTPM 2.0), breaking traditional constraints to deliver end-to-end security and trust for virtual machines. It facilitates compliance with Cybersecurity Classified Protection 2.0 trusted verification and offers simplified cryptographic operations. Supporting BitLocker full-disk encryption for Windows, trusted computing is no longer limited by physical chip count or media constraints, strengthening enterprises’ data encryption, security protection, and compliance capabilities.
Contact Us
