Desktop Reality

Blog archive

4 Pain-Free Ways You can Develop, Execute HA for Virtual Desktops

While the advantages of VDI are clear--from greater user mobility to more agile and efficient user management--many organizations, especially small and midsize businesses, are wary of the significant upfront infrastructure investment and the complexity of traditional enterprise-oriented VDI.

In this post, I will discuss a common question related to virtual desktops: Does reducing the cost of virtual desktops have to mean giving up key features like high availability? The answer, in a nutshell, is no. Below, I've outlined several key steps that will help SMBs set up effective, painless high availability for a virtual desktop infrastructure.

No matter how big or small your organization is, you can't afford to take chances with the consistent availability of your VDI desktops. After all, VDI is supposed to improve agility, mobility, and efficiency--not leave large numbers of users at the mercy of a single point of failure. The challenge is to provide high availability at a manageable cost with minimal complexity.

Generally speaking, implementing high availability requires you to create a mirror of your VDI infrastructure. With traditional VDI, and its many moving parts, this means twice the number of connection brokers, twice the number of provisioning servers and twice the number of load-balancers, as well as high-speed interconnects to a shared storage pool (typically SAN)--it adds up fast. For some organizations, especially smaller ones, this sticker shock is reason enough to skip VDI entirely. For others, especially those that start with a simplified proof-of-concept without high availability, the full cost of high availability becomes clear only when they begin the move to production. At that point, they have two choices: grit their teeth and pay up, or skip high availability and hope for the best.

Here's a less painful way to go about it:

1. Shrink your VDI infrastructure
The simplest way to cut the cost and complexity of high availability is to shrink the size of the infrastructure you need to mirror. Rather than depending on a large, complex array of servers, interconnects, and shared storage, a Citrix VDI-in-a-Box virtual appliance lets you handle dynamic desktop provisioning, load balancing, user and desktop management all from a single off-the-shelf server. This architecture eliminates the overhead of the various management servers needed in traditional VDI. Instead, the all-in-one software runs on the same server that would host the virtual desktops. The solution uses a “grid” architecture: As you need more capacity, you simply run VDI-in-a-Box on additional off-the-shelf servers and join them to the grid, which you manage centrally as one entity from the administration console. User data, including documents and user profiles, is contained outside your VDI server or grid on network-attached storage or a Windows file server.

2. Size your HA server
Technically speaking, VDI-in-a-Box handles high availability through an algorithm called N+1, where N is the number of servers in your VDI grid, and 1 is the number of additional servers you'll need for high availability. This additional server should have capacity equal to the largest server in your grid. If you only have one VDI server, another of similar size will be fine. If you have two boxes in your grid, one supporting 75 users and the other supporting 50, your HA server should be the same size as the larger of the two. This server is a hot-standby, so it is not kept idle during normal operations but serves to offset load and improve performance of the grid.

3. Add another box
Once you've sized your HA server, just install an off-the-shelf server with this capacity--even a spare that's already sitting around in your office--and point it at your primary VDI server or grid. The software automatically creates a high availability grid. If a primary server fails, the HA server steps in, drawing on the same NAS or Windows server to provide a seamless experience for users.

4. Scale at will
As you add servers to scale your VDI-in-a-Box grid, make sure that your HA server is always as large as the largest one in the grid.

High availability should never be seen as a luxury--it's essential to ensure consistent support of your VDI users. This approach lets you implement high availability easily, at a low cost, for uninterrupted productivity and peace of mind.

We hope you've found this guide useful. In our next installment, we'll talk about how you can bring the cost of VDI below the cost of PCs.

Posted by Krishna Subramanian on 01/09/2012 at 12:47 PM


Subscribe on YouTube