chris - Fotolia
Microsoft Storage Spaces Direct simplifies converged, HCI storage
Storage Spaces Direct in Windows Server packages several different Microsoft technologies to lower TCO and simplify storage in converged and hyper-converged deployments.
Microsoft Storage Spaces Direct is a software-defined storage technology introduced in Windows Server 2016 to simplify storage deployment and management. It uses industry-standard servers with locally attached drives to provide storage that is highly available and scalable -- without the expense of traditional storage-area networks or network-attached storage.
Storage Spaces Direct is an evolution of Microsoft's Storage Spaces technology, which was introduced in Windows Server 2012. Microsoft developed software-based Storage Spaces to protect data from drive failures by grouping three or more drives together into a common resource pool, which can then store multiple copies of the data.
Microsoft Storage Spaces Direct uses Storage Spaces as part of its foundation, but also includes other important components --such as Software Storage Bus, Resilient File System, clustered shared volumes, and failover clustering -- to create a complete software-defined storage (SDS) package.
Storage Spaces Direct groups physical drives into a logical resource pool, creating a sharable storage repository that uses the drives more efficiently than more traditional approaches. In addition, it employs a number of technologies to boost performance and protect data, such as caching, erasure coding, storage tiering, NVMe and Remote Direct Memory Access.
Microsoft provides several options for managing Storage Spaces Direct, including Windows PowerShell, Server Manager and Failover Cluster Manager, as well as System Center Virtual Machine Manager and Operations Manager. With the release of Windows Server 2019, Microsoft also added support for Storage Spaces Direct to Windows Admin Center, which helps simplify Storage Spaces Direct management even more.
Deploying Storage Spaces Direct
Microsoft designed Storage Spaces Direct for two deployment scenarios: converged and hyper-converged infrastructure (HCI). The following figure shows the two deployment modes side by side, with Storage Spaces Direct supplying the software-defined storage.
In the converged mode -- also referred to as the disaggregated mode -- compute and storage resources reside in separate clusters, which is important for supporting large-scale workloads. Microsoft Storage Spaces Direct runs on the storage nodes in a scale-out file server cluster to create the storage tier, and Hyper-V runs on the compute nodes in a separate cluster to create the compute tier. The converged mode also uses the SMB 3.0 protocol to facilitate communications between the two clusters.
In the hyper-converged mode, compute and storage resources reside in the same cluster, with storage attached directly to each server in the cluster to create complete but connected compute/storage nodes. The hyper-converged mode tends to target SMBs or remote office/branch office deployments, providing a smaller hardware footprint and lower administrative overhead.
Several vendors now offer HCI based on Storage Spaces Direct, making it possible to purchase a complete HCI that you can quickly and easily deploy in the data center. Dell EMC, for example, sells a number of HCI configurations made up of Storage Spaces Direct Ready Nodes, which are built on its PowerEdge servers. Dell EMC customers can also use Storage Spaces Direct and Hyper-V to build their own hyper-converged systems, both of which are included in the Windows Server 2016 and 2019 Datacenter editions.
Storage Spaces Direct benefits
When Microsoft introduced Storage Spaces Direct in Windows Server 2016, the new feature offered a number of advantages. One of the most notable was the simplicity with which you could deploy it. According to Microsoft, you could configure a Storage Spaces Direct cluster in under 15 minutes. In addition, Storage Spaces Direct could automatically onboard new drives and immediately use them in the storage pool.
Storage Spaces Direct also offered better resource utilization, in part because the erasure coding feature included such capabilities as Local Reconstruction Codes and real-time tier optimization, making it possible to achieve 2.4 times greater resource efficiency. Plus, Storage Spaces Direct could scale up to 16 servers and over 400 drives to provide up to 1 petabyte (PB) of storage per cluster. At the same time, a Storage Spaces Direct deployment -- whether all-flash or hybrid -- could deliver over 150,000 mixed 4K random IOPS per server.
With the release of Windows Server 2019, Microsoft improved Storage Spaces Direct even further. For example, Storage Spaces Direct now supports nested resiliency, which makes it possible for a two-node cluster to withstand multiple hardware failures. In addition, you can now create a storage pool that supports up to 4 PB per cluster of raw capacity. Storage Spaces Direct also added mirror-accelerated parity for creating volumes that are part mirror and part parity, and it supports Intel Optane DC persistent memory modules to achieve even greater performance.
New storage resync alerts make it possible for Windows Server's Health Service to throw a fault message when storage is resyncing. The alerts can help avoid scenarios such as accidently taking down a server during the storage resyncing process. In addition, Storage Spaces Direct now automatically collects performance history and stores it on the cluster for up to one year. This includes data about physical disks, network adapters, clusters, cluster nodes, VMs virtual hard disks and volumes.
A word of caution
Microsoft Storage Spaces Direct lets you use SDS to create logical storage volumes using locally attached storage. With Storage Spaces Direct, you can implement both converged and hyper-converged infrastructures using commodity hardware, while eliminating the need for traditional SAN and NAS storage. There's a caveat to these benefits, however.
Unfortunately, Microsoft only includes Storage Spaces Direct with the Windows Server 2016 and 2019 Datacenter editions, which are much costlier to license than other Windows Server editions. Organizations should weigh their options carefully and come up with an accurate TCO before leaping on the Datacenter edition.
That said, those who have already committed to Windows Server Datacenter have little to lose by at least taking Storage Spaces Direct for a test drive.