Top Gun Project: Week #1

August 8, 2009

Phew! What a week! The main challenge was in figuring out what the project should be and trying to reach common ground with the various stakeholders. We can categorise the stakeholders into three groups:

  1. The Customer,
  2. The Account team, and
  3. The Top Gun team.

Each stakeholder has their own, seemingly conflicting, requirements. The customer, client X, wants to have some form of virtualization in order to obtain operational cost savings. The account team, represented by the CDE, Mr. Y, feels that virtualization requires too much of an up-front cost in investment and that the returns on investment may not be realized. His idea is to have a simple thick-client build that incorporates Intel VPro functionality and that is distributed using Microsoft SCCM. The Top Gun Program leader, Mr. Z, wants something that will demonstrate thought leadership and show value added to the client from a strategic perspective. My task is to marry the business-as-usual cost reduction (value that would be added to satisfy Mr. Y) with the future TCO reductions that come from strategic planning and positioning (added value that would satisfy Mr. Z).

My idea: do it all! I know that, in most cases, one satisfies nobody by trying to satisfy everybody. But I believe there is enough synergy between the desires of the three stakeholders that I can have a phased project: Phase 1 would provide the base Intel VPro build using SCCM (this would be required anyway to do the other phases) and to deliver a server-as-a-service platform, Phase 2 would be to have a diskless client (thin-client) that would operate in a software-as-a-service environment whereby application streaming is used to provide application virtualization to the cloud, and Phase 3 would be to provide instrumentation and predictive analysis tools to understand the future cost footprint of the server-as-a-service platform that uses software-as-a-service. Combining these two SaaS paradigms would lead us to a whole new paradigm: computing-as-a-service (CaaS).

This is what the FSA (Future State Architecture) would look like with the CaaS paradigm:

I fell into a trap in scoping the project. Everyone was trying to assess what the project ought to be by considering the features of the Intel VPro technology. I joined in by suggesting that we ought to be considering the main use cases and assessing what the cost reductions would for Client X for each of those use cases and then use that to decide what to include in the project. I was wrong! We were all putting the cart before the horse. And it is only when I started thinking about what to write in this blog post that it occurred to me that we ought to look at the FSA first, and then decide from the client’s perspective what technologies and processes will add value. (This blog can be a good quality control tool!) So, in essence, we should be thinking from backwards to forwards in terms of assessing the project scope: from the FSA to what we ought to do now. In that spirit, let’s look at the three phases back-to-front:

Phase 3

The biggest problem, in my opinion, with a cloud computing model or an X-as-a-Service (replace the X with software, server or computing) model is how to price the client’s usage of computational resources. Do we charge a flat fee per user on a monthly basis? If so, there is a danger that we may charge the client unfairly because a sizeable percentage of client users may not be using the applications or resources even though we may be charging them for it. This can be bad for business for HP in the long run. We could, instead, charge on a utility pricing model: pay as you use the service. If so, then do we charge on the basis of bandwidth used per user, the time the service is used or the type of application or service that is being used? Considering that our CaaS paradigm is an amalgam of software-as-a-service and server-as-a-service, I think that the pricing should also be a combination of the application or service used and the time over which the computational resources are used. That way, there is a mapping between the two components of CaaS and the utility pricing model’s components. We then add value by analyzing the usage trends and creating a prediction, using an appropriate algorithm, to let ourselves and Client X know what the EUC estate costs will be over the next week, month or quarter. That can be powerful and useful information to the client, I think. Indeed, if I were the client CIO, I would be over the moon if I had that kind of information.

Phase 2

The idea behind Phase 2 is to create a virtualized, disk-less, workstation build at the client end and to provide applications from a private cloud using application streaming. Indeed, Phase 3 assumes that application virtualization is used at the client end, which may be a thick or a thin client. Intel VPro will be used to identify the user and relate his identity to the applications he uses. We could use Microsoft Azure as a platform for the cloud.

Phase 1

Using a Windows 7 build and Intel VPro technology, Phase 1 would be to create a build that would be distributed via SCCM and monitored using SCOM in order to enhance security, track and maintain assets, and create a standardised framework for future builds that would utilize VPro technology.

Conclusion

Finally, I now have a project that just might please everyone, including myself ;-). The downside is that its deliverables have increased whereas the timeline of three months still stands. This will need to be managed carefully.

The thumbprint of the project is given by the table below.

Project Name: EUC Build Incorporating Intel VPro and Analytics for Provisioning Cloud Resources.
Client: Client X has 50,188 desktops and 39,696 laptops, of which 75% are monitored using Tivoli and the others are monitored with SMS. The different processes and tools used for monitoring and the support knowledgebase required for this creates inefficiencies in cost and manpower. Having a single, composite build is required as a matter of urgency in order to address these inefficiencies.
Solution: We propose the creation of a new HP build that has the following features: a) support for Intel VPro monitoring for asset management, tracking and security, b) analytic tools for assessing and predicting the usage trends of computing resources in the cloud (this will help predict the cost of the resources on a utility-based pricing model), and c) Application virtualization to the cloud using Citrix or VMWare.
Value: Monitoring and support costs will be reduced by $100 per annum per workstation, the immediate savings shall be approx. $ 9 million per annum. But over and above this, there should be considerable value added as a result of the cost and usage projections that we could make for the workstations.

The next step, over the coming two weeks, is to sell this project scope to the three stakeholders and to enlist their support.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: