Application Virtualisation Deployment

Continuing with the Virtualisation theme I ran the past month, I am delighted to have a guest contributor sharing some learnings coming out of an 'Application Virtualisation' deployment. Laurent Bouchery is a senior consultant at Microsoft Ireland's Consulting Services, and has been deeply involved in many virtualisation projects over the past year. If you find this article helpful, keep an eye on the Microsoft Consultancy Services Team Blog for more useful tips and tricks from the field.

 

 

'Application Virtualisation Deployment' Article

I had the opportunity recently to work one week for a customer in the North of Ireland who wanted some advice and best practices about 'Virtualisation application', and more specifically about deploying a virtualised Microsoft Office environment. This assignment gave me a new-found appreciation for this technology and for its ability to make your life easier and more exciting. Can you imagine taking an average of just 1 to 4 hours to 'sequence' any moderately complex type applications (and basically 'sequencing' an application means 'packaging' it, having it ready for deployment), as opposed to the days or weeks required for packaging similar applications with the appropriate tools.

Most of the companies have now started or completed their 'Infrastructure Virtualisation' projects (and by infrastructure I mean any operating systems, networks and storage). It is not hard to see why, the technologies are now mature and there are few projects that are as easily justified to your CFO or CTO. However, the flip side is that 'Applications Virtualisation' projects are still few and far between.

I would love to go into a deep dive about the countless benefits that an Application Virtualisation project would have on your company, but my word count is limited for this article. So what I can do is leave you with some nice and compelling numbers sourced from a great white paper ('Application Virtualization: The Next Frontier') that should rest the case for Application Virtualisation:

An 80% reduction in overall application life cycle costs

A 40% consolidation of servers dedicated for your applications

A 400% increase in user uptime

A 30% reduction in application-related help-desk call volume and time

A 4 to 10 months project payback period

Once you have bought into the concept of Application Virtualisation, the next step is to design the appropriate infrastructure. Lucky you, the Microsoft IPD guides (Infrastructure Planning and Design) have just been updated this September to include a specific design guide for Microsoft Application Virtualization 4.5 ("App-V 4.5", the Microsoft Virtualisation application product, previously known as Microsoft SoftGrid). This version of the product was also announced this September as RTM.

One of the first and most important steps in designing your own Application Virtualisation infrastructure is to define which applications you want to virtualise, and to which locations you want them to be deployed. You have to keep in mind that some applications just cannot be virtualised for technical or business reasons. For example:

The ones that require a device driver

The ones that are very tight to the operating system like Internet Explorer or anti-virus software

The ones that are not supported by their vendors when being virtualised

The ones that don't have any specific licensing agreement when being virtualised

About the location definition, you will have to know the number of users by location and the network bandwidth for each location.

So, for designing the infrastructure that will best answer your needs, you will have then basically three choices of model, that can be combined within your organisation and dedicated for specific locations: the Standalone Model, the Streaming Model, or the Full Infrastructure Model.

(Please note that all these models require a Microsoft Application Virtualisation client to be installed.)

A - The Standalone Model: The standalone model is the simplest model. It only consists of a sequencer (the machine use for sequencing the applications), with no additional App-V infrastructure (like a streaming server for example). This model allows virtual applications to be MSI-enabled for distribution, without any streaming capabilities or considerations ('application streaming' is the term used to describe the process of obtaining content from a sequenced application package, starting with minimum mandatory blocks, and then obtaining additional blocks as needed, which allows the application to be quickly available on demand by the user and be cached locally). This model could be useful for remote locations (disconnected or with low network bandwidth) and can be used when the company has already a method for publishing the applications to clients (Microsoft SMS, Active Directory through Group Policy Objects, CDs distribution...).

B - The Streaming Model: The Streaming model is a medium model. It consists of a sequencer, and one or more streaming servers, without specific infrastructure requirements (like Active Directory or a SQL database). As with the previous model, applications are sequenced using the App-V sequencer. However, instead of being packaged with an MSI file, the App-V-enabled applications are placed on a streaming server, which can be a file server, an IIS server, or System Center Application Virtualization Streaming Server. This model is recommended for companies that want to use streaming capabilities but might not have or want advanced infrastructure management. Please note that they can still increase the capability and capacity of this model by using additional technologies and tools like clustering, load-balancing, DFS or SCOM 2007 (System Center Operations Manager can be used to publish and deploy streaming applications).

C - The Full Infrastructure Model: This model is the more advanced model. Additionally to a sequencer and one or more streaming servers, it consists of one or more Microsoft System Center Application Virtualization Management Servers as the core of the App-V system architecture, but also requires Active Directory (for user authentication and application security management), one or more Microsoft SQL servers (for App-V configuration data), and a server running the App-V Management Server Service (for management and reporting purposes). This model is recommended for companies that want advanced features for applications publishing, licences management, reporting and security. Please note also that some fault tolerance and scaling considerations can be increased as it is with the previous model.

To summarise, if you already have a method for publishing applications to clients, then the Standalone or Streaming Models may be enough, and you consider then if you want or need streaming capabilities. If no infrastructure is in place to publish applications or if the advanced management features provided by the App-V Management Server sound interesting to you, then the Full Infrastructure Model should be your choice.

The additional questions that you could have in mind regarding your infrastructure installation could be relative to your clients types (the classic App-V desktop client or the App-V Terminal Services Client), or the types of machine that you will install your sequencer on (physical machines or virtual machines), but they will not affect your design model.

So did you pick your design or combination of models?

I will now leave you in peace to build it. Have fun!

Laurent Bouchery, senior consultant at Microsoft Consulting Services, Ireland