The following is for my fellow Star Wars fans: Remember the first time you sat down to watch Episode I: The Phantom Menace? Episode I, the much-anticipated prequel to the beloved Star Wars trilogy, started strong. Our heroes, a young Obi-Wan and his Jedi Master light saber their way through an army of killer robots, and we imagine what excitement lies ahead as they escape to the planet surface.
Abruptly, our perception changes as the duo run headlong into the slapstick CGI Jar Jar Binks. Suddenly, it is as if we have started watching another movie. Jar Jar frustrates, rather than entertains, and the whole mess devolves into a bad dream from which we wish we could just wake up.
In many ways, Episode I serves as an allegory for the IT world’s experience with virtual desktops. Our young hero isVMware, fresh from the triumph of many server virtualization projects and ready for the sequel.
VDI initially generated considerable excitement. I remember my first exposure to stateful VDI several years ago when the late Clint Battersby (former CTO at Desktone) wowed me while he suspended and resumed his personal desktop session at two different locations in their office suite. I had seen the future, and I was hooked!
Welcome to the future… where my initial euphoria has yielded to the complex and disappointing reality of linked clones. Gone is my personal desktop, replaced by a pale doppelgänger. With linked clones, I am forced to use an approximation of my desktop. They are stateless sessions that restart when I relocate. Worse, to get access to my favorite apps, I must add more layers to virtualize and intermediate applications designed for physical desktops.
The downside of linked clones are many: they are only as good as their master image and so to be efficient, one size should fit most. I have to chuck my existing desktop deployment and management software and purchase and LEARN new software and methodology. Why? Principally, because storage is too expensive to give everyone the 40GB footprint needed by a full VDI. If the VDI storage problem didn’t exist, then neither would linked clones.
Jar Jar was created because Lucas was aiming his film at 11-year-old boys. The actual fans hated Jar Jar because they were 11 years old in 1977. They were ready for something a bit more grown up. In as much as linked clones are designed to satisfy the needs of knowledge workers, they fail to meet the needs of more demanding folks.
The key difference is that we have a choice. Rather than settling for something designed for someone else, we can choose to deploy full virtual desktops. There is only one requirement: inline deduplication.
