vSAN

Just-in-Time Capacity Provisioning with VMware vSAN

A challenge with some traditional storage solutions is the inability to add incremental amounts of capacity. Due to a variety of requirements, policies, and constraints, many organizations will purchase a large quantity of storage capacity once every few years as they attempt to account for current and future demands. Purchasing all of that storage up front means you are most likely paying the highest price per GB. That is because drive capacities increase with newer devices and the cost per GB typically goes down as time moves forward. Purchasing a large amount of storage at once also means you will likely have a significant amount of unused capacity for a while, which still costs money and depreciates over time.

Many organizations allocate funds on a project basis as opposed to a charge-back or “cloud” model. Project-based funding has its own challenges. Example: A new application needs just enough capacity for five virtual machines (VMs). However, the shared storage array is nearly full and there is not enough capacity to support those five VMs. The project manager must now include the cost for a shelf of drives – much more capacity and cost than what is needed for only five VMs.

Hyper-Converged Infrastructure (HCI) solutions make it easier to add incremental amounts of compute resources and storage capacity. This moves us away from the rigid, monolithic patterns of purchasing storage. It takes much of the guesswork out of capacity planning and allows organizations to implement a “just-in-time” provisioning model. In many cases, smaller purchases (a few x86 servers and drives) require fewer signatures, which can speed up the approval, procurement, and deployment of those resources. This means capacity can be added faster  – just in time for that new project. Cost-per-GB is at its best and we do not have inordinate amounts of unused resources wasting away in the data center.

HCI powered by VMware vSAN enables administrators to provision various amounts of incremental capacity for maximum flexibility. With vSAN, it is possible to add a single drive to one host, replace existing drives with larger capacity devices, add or replace several drives, and add or replace one or multiple hosts – all without disrupting existing workloads. Adding resources is simple with just a few clicks. Let’s take a look at a few scenarios…

Here we have a small 3-node all-flash vSAN cluster with running VMs. Each host has one disk group consisting of a cache device and one capacity device. The current total raw capacity of the vSAN datastore is 2.18TB.

An additional flash drive is added to each host. This drive is added to the existing vSAN disk group on each host. Note that this is done without having to migrate VMs or put any hosts into maintenance mode.

As a result of adding an additional capacity device to each host in the cluster, the total raw capacity of the vSAN datastore has increased to 4.37TB.

You can also see in the screen shot above that there is a fourth host that has been deployed (lower left part of the image). We will add this host to the cluster and configure a disk group. As a result, we will increase both the compute resources in the cluster and the capacity of the vSAN datastore.

This next screen shot shows that there are now four hosts in the cluster. It is easy to see which host was added as it is showing 0 of 3 disks in use. It is possible and supported to have a “compute-only” host in a cluster that can access the vSAN datastore, but does not contribute capacity to it. It is recommended that both compute and storage resources are configured and balanced evenly across the cluster. However, this is not strictly required.

Now we will add a disk group for the new host. Each disk group must have exactly one cache device and a minimum of one, maximum of seven capacity devices. The disk group we are creating will be the same as the other three hosts: one cache device and two capacity devices.

Again, we are able to add the host and configure the disk group with no disruption to the workloads running in the cluster on the vSAN datastore. We can see below that the vSAN datastore now has 5.82TB.

HCI powered by vSAN enables organizations to quickly and easily provision just the right amount of compute and storage capacity without disruption to existing workloads. This approach is attractive to a number of accounting models including project-based, show-back, and true charge-back methods. If you would like to take a closer look at how easy it is to add a host to a vSAN cluster, you can click through a demo using this link: Add a Host to a vSAN Cluster

@jhuntervmware