Website under construction

Storage for Hyper-V and Microsoft Azure Stack The primary use cases for Storage 
Spaces Direct is as storage for Hyper-V VMs or as storage for Azure Stack. 

Hardware Storage Spaces Direct makes it possible to build highly available and scalable 
storage solutions using modern storage hardware like SATA SSD for lower cost and NVMe 
SSD for better performance and less CPU overhead. It can also use RDMA-enabled network 
infrastructure for low-latency storage with less CPU overhead than traditional Ethernet. Less 
CPU overhead means increased workload density. 

Prescriptive configurations Microsoft is working closely with its hardware partners to 
define and validate prescriptive server configurations for Storage Spaces Direct. Using these 
server configurations provides the best possible experience with Storage Spaces Direct with 
the full feature set and best performance. 

Storage Configurations You can use Storage Spaces Direct with various storage 
configurations. The most common configurations are: 

SSDs with traditional hard drives, where the SSDs are used as a read/write cache to 
accelerate I/O performance. 

All-flash configuration with NVMe SSDs and SATA SSDs for extremely high I/O 
performance. 

Three tiers of physical storage, NVMe SSDs, SATA SSDs, and traditional hard drives. 

---

55 
CHAPTER 2 | Software-defined datacenter 

Deployment Choice Storage Spaces Direct provides customers with a deployment choice, 
either as hyper-converged infrastructure or as converged infrastructure. In a hyper-
converged infrastructure, compute and storage resources are provided by the same 
machines, simplifying scale and management. In a converged infrastructure, compute 
resources are separate from storage resources, allowing for increased scalability and 
independent scaling of compute and storage. 

Fault Tolerance Storage Spaces Direct is resilient to drive failures. When drives fail, 
degraded data is automatically reconstructed on the remaining drives. Storage Spaces 
Direct supports three fault domain types: a) server, b) chassis, and c) rack, and all data 
placement, data repair, and data rebalancing will adhere to the fault domain configuration. 
