Wanova Redefines Windows Desktop Management and Provisioning

Posted 2010-08-18

Wanova Redefines Windows Desktop Management and Provisioning

By: Frank Ohlhorst dnu

Wanova is offering a new paradigm for managing, supporting and protecting Windows desktops in the enterprise. The company's Mirage product, which entered general availability in March, is designed as an alternative to deploying complex VDI (Virtual Desktop Infrastructure) solutions.

During my testing of Mirage, I found that Mirage offers many of the advantages of VDI (centralized desktop management, application delivery, enhanced support and fast provisioning), without the primary disadvantages–Mirage eliminates the need for connection brokers, hypervisors and display protocols.

That means fewer moving parts, fewer products to integrate and manage and less hardware needed. Since Mirage uses a basic client–server approach and runs the managed desktops on endpoints (desktop PCs), I did not have to set up any server-based computing hardware in my data center; this is where Mirage potentially trumps any VDI solution on the market. It eliminates the need to add processing, high-speed storage and other infrastructure in the data center

Mirage breaks down into three major software components: a server, which stores virtual copies of Windows desktops; a browser-based management console for controlling the virtual desktops; and an intelligent client application, which handles updating and synchronizing the endpoint.

I put Wanova's new desktop management paradigm to the test by installing Mirage on a Windows 2008 R2 server and setting up a half-dozen Windows XP client systems (support for Windows 7 will come in the next version, due before 2010 ends). Mirage costs under $200 per seat, and the company offers volume discounts and other incentives to bring the price point down further.

One of the best ways to tackle what Mirage is all about is to think about the challenges an enterprise network administrator faces daily when supporting user desktops. Those desktop PCs tend to be the weak link in the IT chain and can experience a host of failures, ranging from improperly installed software to improperly patched applications to accidental file deletions. Those problems impact user productivity and can take hours to resolve. Further complicating the situation is distance–those desktops are often located elsewhere, separated by floors, if not thousands of miles.

Mirage addresses those pain points by centralizing the desktop PC–in other words, the user's desktop environment is stored in the data center, where administrators can fully manage it. However, the user's desktop is still executed locally on the user's PC–while this may seem contradictory, it is indeed how Mirage functions. That methodology illustrates the biggest difference between VDI and Mirage–the desktop executes locally and does not require a host in the data center, and users have local control over their desktops.

The secret sauce behind Mirage is the CVD (centralized virtual desktop), which is a combination of a Base Image (BI, or OS layer), which contains the desktop OS, the machine layer (installed applications) and the user layer (user personalization, settings and preferences). Most VDI solutions dispense with the user layer, forcing users to use a cookie-cutter desktop, which they cannot personalize.

The CVD layers can be separately managed by the administrator and then streamed to the remote desktop PC using policies defined by the administrator. The idea here is to decouple the software from the hardware, and then assemble the various layers (often referred to as a layer-cake approach) on the end point to deliver a fully functioning desktop that does not require a hypervisor and executes on the local PC to maximize performance.

Issues of Storage, Bandwidth


Since the primary software components of a user's PC is split into manageable layers, administrators can individually back up those layers and then push the layers down to endpoints to solve problems, similar to a rollback, but with a great deal more granularity and control. At first blush, it would seem that a library of CVDs would require massive amounts of storage (number of desktops multiplied by the storage used on each one) and consume copious amounts of bandwidth (constantly synchronizing endpoint files across the network) when in motion. However, Mirage addresses these issues as well by leveraging de-duplication technology, both on storage and in synchronization traffic.

With the de-duplication technology, hundreds of CVDs may only need to take the equivalent space of a few stored desktop machines. For example, if you have 500 CVDs that are running the same OS and the same applications, you would only need a single copy of the OS and application files, as well as user files that are the same across multiple CVDs. Of course, your results may vary, depending upon the level of commonality on your managed desktops.

On the bandwidth front, efficient compression and an intelligent client application that leverages bandwidth throttling will speed the synchronization process significantly. Because of this, Mirage proves to be an effective solution for remote users connecting over a WAN.

I was able to test both the de-duplication and compression technologies used, and I was impressed with the effectiveness. My six test desktop PCs collectively used about 220GB of disk space; once converted into CVDs, space needed on the network was about 35GB. Moving data over the wire from the desktops to the CVDs and back was very fast, approaching 70 percent of line speed on my 100Mbps Ethernet connections. Further speeding synchronization is how Mirage identifies files that have already been synchronized and skips those files during the synchronization process. No time (or bandwidth) is wasted by synchronizing information that the Mirage server or the desktop already has.

I found the basic deployment of Mirage was surprisingly straightforward and consisted of little more than configuring a Mirage server and deploying the client application to each endpoint. In my experience, that was much less complicated than any VDI product I tested, which required multiple components to be installed, integrated and configured.

The Mirage Client is a small file, and can be distributed via e-mail, downloaded from a Website or easily distributed via other means. Once installed, the local desktop is captured and stored on the Mirage server using a wizard called "Centralize Endpoint." Centralizing an endpoint consists of capturing the desktop contents, including user-installed applications, data and user settings. That can happen in the background, while the user continues to work. Once the desktop was centralized, I was able to create an image of the OS, along with the applications I wanted to provision centrally (for example, MS Office and/or antivirus suites). I was able to store the combination of those layers as a CVD.

I saw the new endpoint listed on the management console under a pending designation, allowing me to execute a wizard to automatically centralize the endpoint. There may be some situations where you don't want to capture every bit of information on a user's desktop. For example, what if the user has gigabytes of MP3s or other non-business-related information on his or her desktop? Arguably, you could create a policy to ban those files or delete them, but that proves to be taking an anti-user stance–not a politically correct move in some organizations.

With Mirage, I created policies that excluded files and directories from the process–I could still let users do what they want on their desktops but not be responsible for allocating storage or backup procedures for non-corporate files. I found this to be an appropriate way to allow workers to use their PCs however they want, yet still protect the core components, applications and data sets needed for their jobs.

Creating CVDs, backing up data/desktops and enforcing policies is only the beginning of Mirage's capabilities. Administrators will find some of the other capabilities incredibly powerful and useful. For example, Mirage allows the creation of a BI from a CVD. That BI consists of the primary OS files needed by a desktop PC. The BI comes from a reference machine (an example system of how an administrator wants a newly deployed PC to look).

If a Notebook Is Lost


The BI can be used to deploy new desktops across the enterprise or automate the deployment of patches or applications. The concept is very much like Microsoft's sysprep process. For example, administrators can create a single BI and distribute it to multiple machines.

Companies that have dozens or hundreds of identical PCs can use a BI to deploy updates, patches, applications, drivers and so on across all machines. The BI is combined with other CVD layers to give the user a customized desktop that features all of their settings, data files and applications. The BI concept will prove to be a powerful tool when it comes to OS updates, such as a wholesale migration to Windows 7 in an enterprise.

A BI can be modified to create different versions. For instance, a BI could be created for a Hewlett-Packard desktop running Windows XP, then another version could be created that offers Windows XP SP2 and another version that includes Microsoft Office and so on. Those versions could be used to repair a damaged OS, roll back a PC to a working state or deploy patches and applications. That bodes well for those providing desktop support.

An interesting addendum to the CVD/BI/Layers approach is the case of a lost notebook computer. In this situation, a traveling executive loses his or her notebook PC–normally, this would be a monumental disaster. With Mirage, the remote worker could purchase a new notebook PC, install the Mirage client and then have his or her CVD streamed down to the new PC. The process works by the Mirage server identifying the minimal set of files that must be transferred to get the remote PC to boot (a significant subset of the amount of data in the full CVD). This allows the remote machine to be up and running in as little as 15 minutes, depending on bandwidth, of course. Once the PC boots, the user can begin working, while additional information and applications are streamed down to the system.

The same concept can be used for a hardware refresh, such as moving from one manufacturer's PC to another. The process would commence with simply replacing the BI and then downloading the CVD to the new hardware.

Mirage delivers further value to administrators by creating a complete inventory of managed machines, which includes details such as CPU, disk size and so on. The product also captures local machine logs, event logs and audit logs, allowing administrators to trace what has happened to troubleshoot machines.

The concept and components behind Mirage may take a little getting used to; the product's underpinnings are quite complex. However, that complexity is hidden from administrators, who now have only to worry about capturing desktop PCs and then using an intuitive dashboard to manage those PCs.

Perhaps it is better to think of Mirage as a platform for desktop management, which stores Windows desktops in a central location and then handles distributing those desktop images to endpoints. Admittedly, that is an oversimplification of what Mirage does, but it does establish the product's goal. To fully grasp what Mirage does, it takes a rethinking of desktop management and especially an understanding of what Mirage is not. Mirage is not a VDI nor is it a traditional system management/software distribution product. Mirage differs from those platforms in two important ways–it does not require a hypervisor and centralizes the desktop in the data center. Once I was able to differentiate between the Mirage way of doing things and other desktop management products, it became a bit easier to give Mirage a chance to prove what it can do in a busy enterprise, one where administrators have to manage and support dozens, if not hundreds or thousands, of PCs.

For users, the management is not at all intrusive. The only difference a user will recognize is improved support and the elimination of manual backup and updating chores. Plus, users are able to personalize their PCs, install applications and retain those changes over time, even if the PCs must be re-imaged or replaced. Mirage proves to be a win-win solution for users and administrators, It is easy to deploy, sensitive to PC, network, server and storage resources, and doesn't require users to change the way they work.


Rocket Fuel