What is the Right VDI Architecture for ROBO?
The VDI 1.0 bandwagon was all about centralizing the management of desktops using virtualization and enabling remote access to desktops for ensuring better security and compliance. Over the last decade, Citrix XenDesktop and VMware Horizon products have established a multi-billion dollar market selling centralized desktop management in the datacenter. Most VDI pilots started out with a small set of desktops and old storage appliances could handle the early load. But as companies scaled, storage invariably became the root cause for performance problems and this caused a major slowdown for VDI adoption across the industry. Almost every storage vendor in the industry today has optimized the stack for addressing the “VDI IOPs” battle and therefore we have seen steady gains in VDI market penetration over the last few years. The price-performance barrier is becoming compelling as OpEx savings are conspicuous for most CIOs.
There are over 11 million branch offices worldwide. Most global organizations have majority of employees working outside the main business headquarters. However, most enterprises have failed to deliver VDI to branch offices. They have not been able to simplify the IT operations in branch offices and VDI pilots have failed despite a lot of promise.
VDI 1.0 network topology for ROBO
There are 3 main reasons for failed VDI 1.0 adoption for ROBO deployments:
- WAN reliability concerns: end to end WAN reliability concerns and cost makes it hard to centralize all the desktops
- Poor performance over high latency links: network latency over 100ms makes it difficult to provide consistent performance to end users. WAN optimization appliances couldn't solve this problem either.
- Complexity: expansion of the VDI management stack from a single product to a unwieldy suite of components has resulted in a lot of complexity that defeats the original OpEx promise.
Here is another way of looking at VDI. We are calling it VDI 2.0.
VDI 2.0 network topology for ROBO
The key characteristics of a VDI 2.0 solution are:
- Only the virtual desktops should be hosted in the branch office.
- Centrally* - All the VDI management should be centralized/SaaS’ified. This includes all the components: connection brokering, image management, load balancers, storefront or app store and workspace logic amongst other things.
- Infrastructure management should be simplified. We only see hyper-converged vendors providing the right building blocks for simplified management.
- The solution needs to be elastic without any capacity limitations both at the infrastructure and management layers.
This is not possible with existing products in the market. For instance, the incumbent product in the market today requires months of work just to configure the network. Check this.
A picture is worth a thousand words!
Here is the proposed VDI 2.0 architecture. A “cloud connector” is co-hosted with converged infrastructure in every branch office for handling provisioning and infrastructure operations. All the VDI management tasks and connection brokering is handled in the cloud. Existing security policies on the network and infrastructure are embraced and no rip-and-replace is done for modernizing the VDI solution.
VDI 2.0 Port Diagram
VDI deployment is simplified because the “management overhead” is reduced considerably and branch office deployment and management is still centralized. This applies to all deployments and is not constrained to ROBO architectures.
As someone with more wisdom once said - “Simple can be harder than complex!”
If you'd like further information on implementation of this solution, please download our free white paper by clicking on the image below.