An Empirical Observation of a “Cloud” Initiative

For a good bit of this past month I have been working with a small business who moved their IT function into “The Cloud”. Obviously, anything cloud-related is a cool thing to talk about, to sell, and to buy. Most definitely, the theory and the promise of a cloud-based infrastructure has much merit: there is a massive migration of Capex into Opex; no onsite server infrastructure means no hardware to maintain, no server rooms, no cooling and power hardware, no computer operations and server administration personnel, and no onsite IT worry; and the company can furnish older or cheaper PCs or the employees can bring in their own devices.

This will most likely be the ubiquitous and blissful model in 3-5 years, but at the moment, the workflows and culture are lagging far behind that technology. Much too much of the IT infrastructure has been assimilated into the marketing hype by vendors looking to capitalize. Indeed, much of e-commerce applications have moved into a cloud paradigm and has reached a level of maturity, but there remains an extremely large base of “legacy” applications and functions that have not migrated. I quote “legacy” to represent the term loosely, since that moniker is frequently meant to degrade perfectly acceptable and extremely successful models of computing.

Any organization considering a solution which entails moving their IT environment into the cloud should also thoughtfully analyze the architecture of their preferred applications, their workflows, and their personal computing preferences. Consulting engagements, books and seminars can, have been, and are continuously being put forth on this subject. Briefly put:

  1. Preferred applications:  The true, or at least, original cloud concept is based on the browser client in a SaaS delivery model. Once the application requires a fat-client or really even vertical browser add-ins, it has left the reservation.
  2. Workflows: Regardless of whether the application is cloud-proofed, if your workflow requires significant onpremise physical interaction, you have potential for problems. For example, if you routinely print 20+ page contracts or brochures, have requirements for USB or (God-forbid) parallel-port connected devices to end-stations, watch out, for you will have bandwidth and end-user support challenges..
  3. End-user preferences: We love our “user-interface (UI) experience”. Change is painful and in some cases, intolerable. Moving from the DOS, Windows 3.1, Windows 2000, Windows XP and Windows 7 desktop “habits” can be unpleasant. Windows Terminal Services, Citrix, and VDI technologies all offer different presentations of the UI experience, but the operative word is “different”. The user population has to be retrained and re-energized.

The implementation that I have just completed is more or less a “private cloud”. Multiple “public cloud” solutions offer alternatives with DaaS or “desktops as a service”, and Network World covered some of those last month. But the challenges remain the same, and those public cloud solutions offer even  less tolerance.