Thu noted that there are two common types of client virtualization technologies:
* Application virtualization--also known as application streaming, it allows specific apps to run isolated from the client device's OS, neutralizing any incompatible issues between the apps and OS.
* Desktop virtualization--also known as virtual desktop infrastructure (VDI), through a hypervisor, it bring the entire desktop OS, processing power and app hosted at the data center. By connecting to the Internet (or WAN), users are accessing an image of the virtual desktop and applications through a dumb client terminal.
Wong added very often enterprises create a hybrid model by using a mix of these technologies to fit different needs within the organization. This is the model at MPI.
Apps are divided into three categories at MPI: a) basic apps, like Microsoft Office, that are compatible with Windows 7; b) apps that incompatible but still widely used among students; and c) apps used in specific academic programs, like SPSS and video editing applications.
MPI's Chan said basic apps are installed in all the PCs. The second type of
applications are being available to students and faculties through VDI technology. The last type of applications, which require heavy computing power, are made available through application virtualization.
"Our ultimate goal is to convert desktop into a utility, like any office supply," said Chan. "Each user will have their personal desktop image and they can access this image whether they are on or off campus."
Surf through the application jungle
Nevertheless, providing quality user experiences with high performance and stable availability was "extremely complicated" for Raths at Swiss Re. "It was very difficult to deal with the (technical) challenges," he said. "If we rushed into virtualization, it could be a problem."
One of the major complications for Swiss Re is its application architecture, which consists of 600 different applications. Not only are these applications often needed by users, many also don't have a Web solution. On top of that, many of these apps are running interdependently or required plug-ins to work smoothly.
Streaming the additional components is a major issue, said Raths, they are creating a bottleneck for the application performance. "In short, you need to have a good understanding of all your applications' interdependencies," he noted.
Apart from the application architecture, the data center infrastructure also needs to be ready to support the technology. "You must be prepared to pay a lot for the storage for each individual desktop image," said Raths. "Imagine if you have 10,000 desktops and each of them require on average 50GB for the desktop image."
Chan echoed the significance of an efficient storage infrastructure, as data centers consolidate the storage burden from each user and PC. "We have adopted storage virtualization and it is a critical foundation," he said.
Sign up for Computerworld eNewsletters.