It can be confusing to consider the many options available for application and workstation virtualization. VDI is yet another common virtualization technology option that would be important to take a close look at if your enterprise has any need for desktop virtualization. If you were an early adopter of VDI then you may have quickly learned that this option doesn’t scale well on traditional architecture, but with some of the technology advancements of today this a really great option to be researching.
Here we will take a look at some of the compelling reasons to use VDI, then take a look at important technical and cost considerations. From there we will dive into some monitoring aspects that are extremely valuable to a successful and stable deployment. Let’s get started!
When should I use VDI?
The answer to this question will be different for every organization, so it’s important to take a close look at this from all perspectives. The following table highlights both the business benefits and the time saving technical benefits that I commonly see for decision making around this technology initiative.
In reviewing this list, I suspect that you made a connection with many of the benefits above that would be suitable for your organization. Keep in mind that VDI may not align with all your virtualization needs. It is completely ok to only use VDI for the business use cases that provide the most business benefit for your enterprise users.
When VDI was first introduced, the idea of cost savings with VDI was null and void unless you were considering this technology strictly from the security perspective in that it was easier to push updates and patches to your virtual workstations. Fast forward to today and the infrastructure required to deploy VDI is no longer cost prohibitive, the performance is now there, and organizations are more prepared to embrace this type of deployment.
Some of the technology advancements that make VDI a solid choice for workstation replacement are as follows:
- Storage advancements – Many storage vendors now have options specifically designed for VDI to ensure that VDI workstations get the I/O required for a great user experience for VDI. Take a look at hyperconverged options, flash storage, and any similar combination of storage offerings to be successful with VDI. If you are not in a position to buy new storage for your VDI deployment then take a look at some of the software-defined product options out there (such as VMware vSan or similar) that allow you to use existing hardware in a way that can provide the performance needed for a successful deployment.
- GPUs – Everything a user does on a workstation has some level of graphics intensity. To ensure your virtual workstation doesn’t get sluggish when rendering graphics be sure to test and deploy GPU’s. Some will even argue that you should deploy up to 1 GPU per user, but I would encourage you to test with your organizational applications before finalizing on that decision. Not all applications require access to an entire GPU to run successfully.
Next let’s look at a couple of the deployment options that can improve the VDI experience for the administrator. These tools become essential to the administrator management experience of your deployment, and are key to your deployments success from a management perspective.
- Machine Creation Services (MCS) – Is a deployment tool that essentially allows you to take a copy of a virtual machine by communicating with the hypervisor, and allows you to deploy clones of the machine for business user workstations. The stability of this option relies heavily on your storage. One of the really great things about MCS is that it works with XenServer, Hyper-V, vSphere, but also Microsoft Azure and Amazon AWS, making this option more versatile for the future.
- Provisioning Services (PVS) – Is a deployment tool that allows you to create images and deploy them to business users as workstations. This option relies on network streaming of that image, and can deliver to virtual or physical endpoints. The stability of this option relies heavily on your network. This option tends to be better for image lifecycle, and update deployment to your images and the enterprise.
What Types of Cost Savings Can I expect?
The cost savings for VDI come from several factors. For one, administrators can save time managing and deploying desktops allowing IT teams to run in the leaner fashion that organizations of today prefer. Another cost savings component involves more secure workstations through improved process for rolling out patches to organizational workstations. Ensuring that workstations are available when needed for patching, because they are secured in the datacenter ensuring your workstations the opportunity to be more security compliant than ever. Lastly the flexibility to run desktops from any device making VDI a workstation replacement. Allowing corporate owned physical PCs to need less compute over time meaning that they can be kept longer or even eliminated if the business user is allowed the option to bring in their own device to work.
VDI offers great flexibility and security to the enterprise which are proving to be more and more important to enterprises today.
Comparison (VDI vs Standalone PC)
Analyzing the requirements for VDI are slightly different than the other analyses we have done for some of our other previously covered topics such as Published Applications and RDSH. VDI is deployed directly to the hypervisor, so you are really sizing hosts for these deployments instead of individual virtual servers. So, I have created a table that highlights these key differences for your review.
In this comparison you can see that the resources on the host can be shared with all the virtual desktops you deploy. This allows for the ability to consume most if not all the resources you purchased instead of enduring the waste of unused resources that comes with purchasing individual PCs, as demonstrated above.
Keep in mind that with VDI if you choose to leverage PVS for deploying and scaling your workstation images you will need to plan for a PVS server installation. PVS servers can be physical or virtual. The average PVS install involves at least 2 PVS servers for redundancy, sometimes more. The number of images per organization will vary, but let me give you an example of an enterprise that has 15 different use cases for VDI. In this case they have 15 images that support 150 virtual workstations, and their deployment was most successful when each PVS server had 8 vCPUs and 64 GB of memory.
Monitoring Challenges to be prepared for
VDI is a great technology to use when you are looking for an option that will deliver a virtual workstation that performs the same as, or even better than, the traditional workstation. That does mean that there are many moving parts, and some complexity that is a little bit different than other virtualization options. The differences make it even more important to monitor your environment, but it also means that the tool you choose must be able to monitor every aspect of VDI.
Let me explain by taking an example of a healthcare organization that uses virtual workstations delivered to thin clients they have at the patient bedside. This organization has about 1,000 beds, so that means that they are deploying 1,000 VDI workstations to be prepared for maximum capacity. Now consider that fact that this deployment also includes the use of PVS for scaling up and scaling down workstations, and it leverages GPUs to ensure that any radiology images can be rendered on these VDI workstations. That’s not all – they have chosen a VMWare hypervisor with Citrix XenDesktop software for the control plane and management platform.
Monitoring an environment with a design as described in this real-world scenario is complex, and to take on the task manually would be a challenge. It would be a very tedious, error-prone process which is counterproductive and a waste of valuable time.
To monitor your VDI environment successfully you will need a robust solution. When choosing your monitoring tool, it will be important to ensure that the following metrics are available for successful monitoring of VDI.
- Ability to monitor the storage controller
- Write cache insights
- Physical memory consumed by the host
- Virtual memory used by the VDI workstation
- CPU and GPU monitoring with the latter being more important
- PVS and MCS insights including storage monitoring
- Visibility into the inner communications and functionality of all virtualization infrastructure deployed
- Hypervisor monitoring
The value of monitoring your environment is very important and the above will provide a good foundation of the many important aspects of monitoring your deployment successfully.
What will you do?
Whether to use VDI can be a tough decision, but it may be the best option for business alignment and business user benefits for the long-term. For example, if the business is looking for their teams to be able to work from any device, but get the performance of a physical workstation then this is probably the option for you. Take the time to weigh all the risks and benefits before committing, and then follow through.
If you determine that VDI is indeed the right virtual workstation deployment option for some or all your business use cases, then I would strongly urge you to evaluate and implement a monitoring tool. As we discussed there are many different components to a VDI deployment, and to manually discover an issue with the controller, GPU, or any other component of the environment would be a time-consuming initiative. The right monitoring solution becomes, in a way, your insights into your deployment in a way that you can feel confident that you are proactively aware of the performance of your implementation. The right monitoring tool is not a nice-to-have – it is a necessity.