One of the buzzwords in the IT world for the past couple years has been Hyperconverged Infrastructure or HCI. HCI threatens to disrupt the traditional server and storage area network (SAN) that most institutions have been deploying for the past two or more decades by combining the CPU, RAM, network, and storage resources into the same hosts. Simplivity (now HPE) and Nutanix were some of the first vendors on the scene. When VMware announced vSAN, and later, Microsoft added Storage Spaces Direct (S2D), HCI really took off.
What exactly makes HCI so attractive and “game changing”? One of the key components of a traditional SAN that made them faster than traditional drives was a cache for storing items before actually writing to a hard drive. The cache was much faster than the drive, so performance could be boosted. Once flash storage drives started appearing and costs came down, the flash cached SAN (most notably Nimble, now HPE) started to become popular. These SANs used much larger flash-based storage as a read and write cache. vSAN and Storage Spaces Direct use the same basic principal as the flash cached SAN, but move the storage from dedicated expensive hardware to the same server hardware that is running the hypervisor. Both HCI technologies can be deployed as a hybrid (flash drive for a read/write cache and spinning drives for capacity) or all flash configurations. With the introduction of NVMe flash drives, performance unheard of a couple years ago can be obtained with very little hardware.
Most schools do not have a dedicated storage manager, so ease of use is another benefit for running HCI. In vSAN and S2D the storage is a dynamic pool that grows or shrinks as the environment changes. Both products can scale out (add a new server) or scale up (add more drives to the servers). Newer 2U servers have the ability to house 24 or more 2.5” hard drives, so it is easy to have an environment that can grow its storage considerably, simply by adding additional hard drives to the existing servers. The environment will see the new drives and add them to the total capacity. If a drive fails, the total storage capacity goes down, until it is replaced.
Protection of the virtual servers can be as basic as a two-way mirror, sometimes referred to as RAID 1 (two copies of the VM’s hard drive and a witness) or multiple copies of the VM can be spread across more servers. With more cluster nodes, erasure coding can get you a RAID 5 (less storage) or RAID 6 (more resilient) protection.
When comparing the costs to replace a SAN, HCI can be less expensive, especially when adding the costs of replacing the servers. For small to medium education customers, the server and storage needs for the entire district could be handled by two or three servers.
In future posts, I will discuss IU13’s dive into the HCI pool along with some lessons learned along the way.
Tech Talk Live is the only conference of its kind in the region specifically designed for IT pros in education.
1020 New Holland Avenue
Lancaster, PA 17601