How to Bulletproof Your VDI Deployments
In our previous blogpost, The CYA Approach to VDI, we shared how to minimize problems when deploying a virtual desktop infrastructure. After deploying and supporting storage for hundreds of companies, the Tegile team has gained significant insight about storage for VDI environments. Before we share that insight, we’d like to bust a few myths with regard to legacy storage performance. The following names have been changed to protect the innocent.
My legacy SAN/HDD array can handle 70 servers and 8,000 IOPS. Surely it can handle a couple hundred desktops!”Systems Administrator, Financial Services Company
The performance level of legacy systems does not necessarily carry over into VDI environments. Remember that overall performance is largely determined by the storage system. With insufficient storage performance for a traffic load, user experience will suffer. With insufficient storage, when doing maintenance you’ll also get errors and broken desktops. Correcting broken desktops can keep you working all night long. Surely you have better things to do than spend a couple hundred hours re-setting desktops.
But, I’ve only seen it do 5,000 IOPS when I’m really pushing it with VDI.”
Data Center Manager, National Health Care Institution
Your I/Os are about 90 percent random and inconsistent. Remember that workloads can be “bursty” because of boot/login storms, virus scans, and patch updates. Also remember that I/O bursts can be very write-intensive, based on the workload. We have a user whose IOPS peak around 40,000. Legacy spinning disk systems are the worst for these scenarios, where it’s nearly impossible to serve consistent performance. And what happens when you need to scale?
But I can over provision to cover performance spikes.”
IT Director, Global Retail Company
You can over provision your storage, that is, if you don’t mind costs spiraling out of control. Nearly everyone wants their legacy storage to work in new computing environments. But if it were that easy, we would not be talking about how to bulletproof your VDI deployment. So, let’s hold off on wishful thinking. Instead, let’s review the lessons we’ve learned to successfully deploy all-flash and hybrid storage in VDI environments.
All-Flash or Hybrid?
Now that we’ve busted the myths associated with VDI on legacy storage, let’s discuss the other storage options: all-flash array versus hybrid array. When implemented in a VDI environment, both all-flash and hybrid arrays — and Tegile arrays in particular — offer a number of advantages over legacy storage.
We can all agree that the performance of flash media far exceeds that of traditional spinning media. Tack on the ability of flash media to handle random I/O with flying colors, and you have one heck of a VDI machine, potentially capable of mitigating performance killers, such as boot storms.
But what if the cost of all flash is prohibitive to your project? Hybrid arrays to the rescue! With hybrid arrays, you get all of the benefits of flash performance in the front, with the cost benefits of traditional spinning media in the back. With Tegile hybrid arrays, you also get the benefits of a proven architecture that ensures frequently used data is served from high-speed system cache.
The truth is that all-flash arrays and hybrid arrays are both candidates for VDI deployments, depending on your workload:
- All-flash arrays sacrifice some capacity for extreme performance
- Hybrid arrays offer flash performance, with the higher capacity and lower cost of hard disk drives
Whether you’re looking at all-flash or hybrid storage, capacity and cost of ownership can go further when you include inline deduplication and compression.
A caution about costs
Do not dismiss all-flash storage because it seems expensive. When our customers calculate return on investment of all costs associated with running storage in a datacenter, they find that when flash storage is installed, their cost per gigabyte goes down while workload performance goes up. When consolidating workloads, see which metrics to consider when calculating ROI, as listed in our blogpost, Consolidate THIS.
Remember that vendors supply cost and performance metrics based on their own workloads and measurements, which are probably not the same as your specific workloads and performance needs.
Storage costs depend on multiple factors:
- Number of virtual desktops to support (there’s an economy of scale)
- Number of desktops that are persistent vs. non-persistent
- Performance requirements of each desktop (IOPS)
- The capacity needs of user workloads
- Number of software licenses required
- Features required to support
- The type of maintenance included with the storage purchase
- The rack space required to house storage
- The cooling costs required to maintain storage
Why we rule at VDI
At Tegile, because we offer both all-flash and hybrid storage, we can dial up or down the amount of flash needed for your environment, workload and budget. All Tegile arrays support multiple protocols – FC, iSCSI, NFS and SMBv3 – depending on the user’s specific needs. Features such as thin provisioning, inline deduplication and compression, and data protection make VDI a much less-daunting project to deploy and manage. With Tegile systems, you can expect an increase in performance.
Data reduction and capacity optimization are the secret sauces to VDI. Most desktop users only use a fraction of their total storage capacity. Thin provisioning automatically allocates storage space as data is being written. Any space that hasn’t been consumed remains available.
Using dedupe and compression can shrink a storage footprint up to 90 percent. Duplicate data from OS images, applications, libraries, and user-specific applications that are spread across virtual desktops are automatically identified and reduced to a single instance. Dedupe and compression allows hundreds, or even thousands, of persistent desktops to be shrunk down to one single gold/master image, and likely read directly from high-speed memory.
Data protection functions in Tegile arrays include space-efficient point-in-time snapshots, bandwidth-efficient remote replication capabilities, VM-consistent snapshots for virtual desktop images, and application-consistent snapshots for app data. Users can recover an entire array, one or more VMs, a single desktop, or a single file.
Tegile storage certified for VMware, Citrix and Microsoft
Finally, Tegile arrays are certified for VMware, Citrix and Microsoft. Whatever type of hypervisor and virtual desktops server solution you use, the team at Tegile has already tested it and ensured interoperability. We’ve also created reference architectures you can use as a guide to help you minimize problems when deploying VDI.
As always, if you have questions about how to deploy all-flash or hybrid storage in any computing environment, don’t hesitate to reach out to our team at Tegile. After successfully deploying VDI-focused storage for hundreds of companies, our team is eager to share our insight and help you plan your next storage move. If you want to chat about storage plans for your company, reach out to us at firstname.lastname@example.org