VDI storage requirements
VDI or Virtual Desktop Infrastructure was a revolutionary technology that brought server virtualization into the desktop world. User-facing, fully functional desktops could live as virtual machines with all the ease-of-management benefits offered by virtualization.
While in existence before VMware hit the scene, their success in the market made VDI popular and much more commonplace. Moving from a distributed, physical desktop fleet to VDI has a lot of benefits but there are also a few challenges and considerations that need to be accounted for, including the topic of this article: storage.
Storage for a Virtual Desktop Infrastructure solution comes into play for a few different needs: desktop VMs, user data, high-speed cache. When a full desktop is converted from physical hardware to a virtual machine, all of the hardware component bits are moved into a software layer. This includes memory, operating system, and storage. Storage is further broken into desktop, user data, and high-speed cache. The desktop or VM storage is used for all the individual desktop instances that will run in the environment. User data may be internal to the instances as with traditional assigned desktops or off-loaded to separate storage when using non-persistent desktops. High-speed cache is optional but offers great benefit during recompose/refresh operations and to better handle boot and login storms. While storage is fairly trivial for physical desktops, it’s a critical foundation component when deploying a VDI solution.
When scoping out a VDI project and reviewing the storage needs, there are quite a few factors that need to be considered. Here are some of the most critical and common to include in your project:
Persistent or non-persistent desktops
Are the storage requirements different?
One of the first choices to make with a VDI deployment is around desktop persistence. VDI comes in two flavors: persistent and non-persistent. Persistent desktops are assigned to individual users where they login to the same VM each time, exactly as a user would with an assigned physical laptop or desktop. Non-persistent desktops are basically disposable and are re-created from a template for each new user. This is similar to the behaviour of a computer lab when a product like Deep Freeze is in place. The storage requirements for each mode is different, as the compute profile on the backend infrastructure is also different. Persistent desktops are often running all the time and only logged into at the start of each day. This puts a very predictive load on the storage and therefore a lesser class of more inexpensive storage can be used. On the other hand, non-persistent desktops are very IO intensive and will require higher performing storage and maybe even flash-based cache to meet the performance demands of users.
How will students and staff be using the system?
Another factor for storage considerations is the overall use of the VDI system on a daily basis. Will it be a lot of staff logging in at 8:00am and then logging out at 5:00pm, or will it be a lot of students logging in and out every hour to hour and a half for classes? Login activities can create a high load on storage and should be accounted for properly.
Ensure storage can deliver the expected workloads and prevent user frustration.
Similar to user requirements (above) is in-VM performance. Will users be working with file-laden programs, in-VM databases, or any other workload that relies on disk performance to be high? If so, then a higher tier of VDI storage will be required.
Look at the number of users and desktops required
Consider scalability going forward.
Scalability is another area of consideration when designing your storage needs. How easy is it to expand the storage should capacity needs increase? What if the VDI desktop profile suddenly changes and now needs 25% more storage? Being able to effectively and efficiently scale VDI storage will help with long-term success.
Applications being used and how much storage is needed
In Higher Ed, the number of applications being deployed in the VDI environment is a key factor to consider when selecting VDI storage. Are those applications IO intense? Do they have a large installation footprint made up of thousands of individual files on disk? What kinds of datasets are normally used with each application? All good questions to have answered as part of the design phase of a VDI project.
Predictable user experience
Maintain user experience even during peak load times.
Again, in Higher Ed, class changes can put an immense load on a VDI environment and particularly storage as login storms are common. When hundreds or even thousands of students all try to login to their desktop for the next class, your VDI storage better be able to process those requests without issue.
Costs of different storage solutions
A large infrastructure deployment such as VDI often carries cost considerations as well. The storage component is no different and choices abound. Spinning or flash or hybrid? Single-level cell (SLC) or multi-level cell (MLC) flash? Separate high-speed cache array?
How to support back up and disaster recovery operations
A primary difference and perhaps drawback in moving from a traditional, distributed physical computer fleet to VDI is the introduction of a single point of failure and thus the need for disaster recovery and business continuity (DR/BC) plans and systems. How will you respond to a partial or complete failure of the datacentre bound VDI platform? Can you recover quickly and completely? Had user data and applications been protected?
- Direct attached storage (DAS)
- Network attached storage (NAS)
- Storage area network (SAN)
Direct attached storage (DAS)
Direct-attached storage is exactly as it sounds; digital storage connected and attached directly to the hardware of a physical machine. IT might use DAS drives to provide thin clients with temporary local storage for users to save to, or allow the machine to load programs from a golden or master image.
Network attached storage (NAS)
NAS refers to any digital storage accessible via a network connection. This type of storage is used to provide access to shared drives or deliver applications or resources from servers.
Storage area network (SAN)
Storage area networks are very similar in function to NAS, but appear to users' machines as Direct-attached storage. This allows the machine to treat and interact with SAN storage as such and mirror all the functionality of DAS drives.
- Poor performance
- Poor overall user experience
- Dissatisfied students
As you can see, the storage subsystem for VDI is one of many critical components and when not architected properly can result in poor performance. When the storage layer in VDI suffers, so does the user experience and for a Higher Ed institution, a bad student experience can affect satisfaction ratings, enrolment, and retention.