MayaScale Composable NVMe Storage

MayaScale delivers disaggregated high-performance NVMe storage for Azure by transforming L-series ephemeral NVMe SSDs into persistent, shared storage pools separated from compute instances. Unlike traditional direct-attached storage that creates stranded capacity and deployment restrictions, MayaScale consolidates NVMe resources into flexible shared pools that scale from 400K to 2.88M IOPS with policy-based deployment and transparent per-vCPU pricing.

Disaggregating storage from compute eliminates over-provisioning, enables independent scaling of storage and compute resources, and removes the deployment limitations of locally attached ephemeral NVMe. Storage volumes are accessible via industry-standard NVMe-oF over TCP and iSCSI protocols using native kernel drivers—no agents, proprietary software, or client-side modifications required. Server-side replication provides data persistence and active-active high availability with automatic failover, operating transparently without impacting client performance.

Key Benefits:

• Disaggregated architecture separates storage from compute, eliminating stranded NVMe capacity and enabling flexible resource allocation
• Policy-based deployment automatically selects optimal Azure L-series configurations based on performance requirements, eliminating manual infrastructure sizing
• Disruptive per-vCPU pricing delivers enterprise-grade performance at a fraction of traditional storage costs with no hidden fees or vendor lock-in. 1-month free trial available
• Server-side replication transforms ephemeral NVMe into persistent shared storage with sub-minute failover and sub-millisecond latency
• Dynamic provisioning allows storage resources to be allocated and returned as workload demands change, maximizing utilization
• Works with standard Linux distributions (Rocky Linux 9, Ubuntu 20.04+, RHEL 8+) using native NVMe and iSCSI kernel drivers

Ideal for:

• High-performance databases requiring consistent sub-millisecond latency without over-provisioning storage
• Analytics and big data platforms processing large datasets with intensive I/O requirements
• AI and machine learning training workloads demanding scalable IOPS that match compute resources
• Containerized applications on Azure Kubernetes Service needing persistent volumes with enterprise performance
• Infrastructure teams seeking simplified storage deployment with transparent, predictable pricing

Technical Features:

• 5 configurable performance tiers supporting Azure L-series families: Lasv4, Laosv4, Lsv3, Lsv2
• NVMe-oF (NVMe over TCP) and iSCSI protocol support with standard kernel drivers
• ZFS-based storage management with snapshots, cloning, and compression
• Active-active clustering with automatic failover and load distribution
• Dual-NIC architecture separating client and replication traffic for maximum throughput
• Support for single-node deployments or multi-node HA clusters