VMware vSAN and HCI offer secure shared storage for VMs
Coupling vSAN's all-flash storage features and vSphere availability support with HCI can lead to better resource utilization and workload performance.
VMware administrators looking to streamline virtual server operations, optimize resource utilization and boost performance should consider implementing vSAN HCI.
VMware vSAN provides software-defined storage services for hyper-converged infrastructures (HCIs) built on VMware technologies. VMware vSAN works with the vSphere hypervisor to deliver a virtual storage area network that spans the infrastructure's server nodes, providing secure shared storage for VMs.
Admins can purchase out-of-the-box HCI appliances that incorporate vSAN, such as Dell EMC VxRail, or they can build their own HCI systems using standard x86 servers validated to run vSAN.
In either scenario, vSAN supports both all-flash storage configurations and hybrid configurations that incorporate solid-state drives (SSDs) and hard-disk drives (HDDs). VMware vSAN also offers native encryption for protecting at-rest data and includes features for optimizing all-flash storage, such as compression, deduplication and built-in caching.
The vSAN cluster
Like many HCI platforms, vSAN HCI is made up of multiple server nodes that form an integrated cluster. A vSAN cluster includes two to 64 nodes, so organizations can start out small and add nodes to scale out.
The system configures each node with direct-attached storage drives that vSAN aggregates into a logical storage pool. All VMs share that storage pool, with no VM tied to a specific physical drive.
VMware recommends that each node be a vSAN Ready Node: an x86 server configured, tested and certified for vSAN. A Ready Node server optimizes the processing, memory, networking, controller and storage resources necessary to support vSAN implementation. A number of vendors offer Ready Node servers, including Dell EMC, Fujitsu, Hitachi, HP Inc., NEC and Super Micro Computer.
Unlike some HCI platforms, vSAN HCI doesn't require special software or storage virtual appliances installed on each node, leading to better performance and resource utilization.
VMware vSAN is an object data store that mounts storage devices across the HCI cluster and presents them to vSphere as a file system. Elements such as virtual disks and configuration files are known as objects. Each object is made up of one or more components that the system distributes across the server nodes. The size and number of components depend on object size and applicable storage policies.
Disk group configurations
Each node in a vSAN configuration includes the compute and storage resources necessary to support VM-based workloads. VMware vSphere abstracts the resources and makes them available to VMs, using vSAN for the storage component.
The system divides the storage devices on each node into disk groups. A node can support one to five groups. A group consists of a cache tier and capacity tier. The cache tier includes one device, and the capacity tier includes one to seven devices.
A group can contain all SSDs or a mix of SSDs and HDDs to form a hybrid configuration. VMware vSAN doesn't support HDD-only groups, nor does it support different group types in the same cluster. Admins must set up all nodes with either all-flash or hybrid groups. In addition, the device used for the cache tier must be a flash drive, which means that admins can't use vSAN without at least one SSD on each node.
In an all-flash group, the cache device serves as a write buffer. There's no need to cache reads, because flash drives in the capacity tier have plenty of power to handle read operations. VMware vSAN HCI supports Serial Advanced Technology Attachment, Serial-Attached SCSI (SAS), Peripheral Component Interconnect Express and non-volatile memory express flash devices.
Normally, the cache device in an all-flash group is a lower-capacity, higher-endurance drive, and the capacity devices are higher-capacity, lower endurance drives. VMware vSAN performs writes in the cache tier and then de-stages the data to the capacity tier where it is stored. This configuration can help maintain performance and extend the life of the drives in the capacity tier, leading to lower storage costs.
In a hybrid group, the system uses the cache device for read caching and write buffering, with 70% of the capacity allocated for reads and 30% for writes. As with all-flash groups, data is de-staged from the cache tier to the capacity tier. The capacity tier is made up of all HDD drives, which are either SAS or nearline SAS disks.
Cluster configurations
VMware provides several options for implementing a vSAN HCI cluster. Admins can deploy a standard cluster that includes three to 64 nodes. The standard cluster is commonly implemented in a single physical location on the same Layer 2 network.
Admins can also deploy a two-node cluster. Again, the cluster is usually deployed in a single location and network, although admins can also directly connect the nodes with crossover cables. In addition, a two-node cluster requires a witness host to serve as a tie-breaker in the event that connectivity between the two nodes is lost.
Another option for implementing vSAN is to create a stretched cluster that spans two locations, which can help protect against data loss. In this configuration, admins distribute the nodes across two sites and replicate the data between them. Each site can support up to 30 nodes. A stretched cluster also requires a witness host in a third site to serve as a tie-breaker.
Software-defined storage
Organizations can use vSAN HCI wherever hyper-convergence makes sense, whether private clouds, public clouds, edge environments or centralized data centers. VMware vSAN offers policy-based management, hardware compatibility checks, storage capacity reporting and vCenter diagnostics. VMware vSAN also supports vSphere availability features, such as fault tolerance, high availability and VM replication.
Virtual servers running on VMs in a vSAN environment have full access to the cluster's storage pool, without being limited to the host's local storage. Not only can this lead to better resource utilization, but it can also help better accommodate today's dynamic and fluctuating workloads.