Skip to main content

NAS Storage (Shared File System)

NAS Storage is an on-demand, highly cost-effective shared file system delivered via standard network protocols (NFS/SMB). Designed for modern AI and cloud-native workloads, it allows you to completely decouple your data from your compute.

Whether you are staging massive datasets for HyperTrain, persisting model checkpoints from a Cloud Container Instance (CCI), or sharing custom files across your team, NAS Storage provides elastic scalability and rock-solid data persistence. Spin it up, mount it seamlessly across multiple instances, and ensure your data outlives your compute lifecycle.

🚀 Key Features & Storage Tiers​

Alaya NeW Cloud offers two distinct tiers of NAS Storage to perfectly match your workload and budget:

1️⃣ High-Capacity NAS (Cost-Effective & Versatile)​

Built for general-purpose, long-term data persistence where high capacity and low cost are the priorities.

  • Universal Compatibility: Fully POSIX-compliant. Mount it seamlessly to your CCI, Alaya Lab, or distributed inference tasks.
  • Economical Persistence: Provides long-term, high-durability storage at a fraction of the cost. Ideal for logs, backups, and raw data staging.
  • Secure & Reliable: Features multi-layered security protocols to ensure your data is always safe, isolated, and highly available.

2️⃣ High-Performance NAS (Built for AI & HPC)​

Engineered for I/O-intensive workloads that demand extreme throughput and ultra-low latency.

  • Extreme I/O Power: Drastically accelerates storage I/O, feeding data-hungry GPUs faster to maximize your Compute efficiency.
  • Sub-Millisecond Latency: Powered by a distributed architecture, it delivers near-instant read/write speeds for latency-sensitive applications.
  • Massive Concurrency: effortlessly handles tens of thousands of IOPS, making it the perfect backbone for massive AI training datasets, high-frequency trading, and demanding databases.

🌟 Core Capabilities​

  • Elastic Scalability: Scale your storage volume horizontally on the fly. Pay-as-you-go for what you provision, avoiding idle resource waste.
  • Unified Access Control: Centralized permission management across projects and users. Granularly share access securely across your team.
  • Decoupled Persistence: Teardown your CCI or HyperTrain tasks without fear. Your storage volumes live independently, ensuring zero data loss when compute instances are released.
  • High-Efficiency Access: Built-in real-time monitoring and quota tracking for high-throughput, I/O-intensive workflows.