Automating the Cloud

Enterprise IT Automation: Stuck in Model T Mode
Henry Ford democratized the car, introducing the assembly line into his Model T factory and, thus, igniting a revolution that forever changed manufacturing. By automating parts of the assembly process through the use of a moving conveyor belt, he accelerated the entire process and enabled more cars to be produced faster. This concept of mass production quickly spread to other industries, enabling companies to extend their reach to many more consumers with standard, albeit limited, versions of their products.

These days, most consumer products are becoming democratized, meaning, accessible to everyone. Subsequently consumers expect more personalized products, and in many industries that’s what they get. Automation has evolved to enable manufacturers to automate discrete, commoditized parts of a product – such as the chassis, engine and wheels of a car – and then assemble those standardized parts to create a unique end-user product. In this way, we can choose cars based on a variety of factors, from how we’ll use them (commuting, carpooling families, hauling goods), to what colors and fabrics we prefer, to how much fuel we’re willing to consume.

Sectors of the consumer IT industry have also adopted this evolved automation model. For instance, we can buy computers online from a wide variety of vendors and, using self-service, configure exactly the type of device we want, specifying the amount of RAM, hard drive, battery life, type of operating system, applications, and more.

Unfortunately, this level of personalization has not spread to enterprise IT departments, which remain stuck in an automation paradigm that resembles the original Henry Ford model. As a result, consumers, i.e., corporate employees, are often extremely limited in the type of IT resources they can obtain. For instance, developers or business groups that need a new server, even if it’s a virtual machine (VM), are given just a couple of options in terms of CPU and RAM, to which they must accommodate their needs and their projects. Until now, while this has been a source of much frustration for users, it has not been a significant business impediment. But with more companies expanding their use of virtualization and wanting to adopt cloud computing, it threatens to stifle and even derail enterprise cloud initiatives.

Enterprise IT departments are not to blame for this rigidity nor its implications. Rather, the problem lies with the automation products they’re using. Most either enable very little variation and, therefore, require a lot of human intervention to tailor resources and make them useful for employees, or they require IT to create so many upfront permeations to address disparate needs that cloud-based resources become incredibly unwieldy and cumbersome to manage and support. For example, many products that purport to automate provisioning of virtual resources follow a pipeline approach. They require administrators to hardcode all the parameters related to a VM upfront, such as how much memory a VM will have, where to place it, which storage resource to use, and what price to attach to it. Typically, the options are very limited and the result is thousands of virtual machines that are essentially the same, built with scripted automation that follows the same sequence.

But as democratized manufacturing spreads through our personal lives and virtualization spreads throughout an organization, the configurations that users demand will vary widely, requiring much more individualization than traditional tools are capable of handling. In addition, the relentless march of desktop and server virtualization in the enterprise will lead to much higher churn in the environment. This means that resources that may have been available or that made financial sense to assign when they were first requested, may no longer be available or feasible when the requests are processed a few hours later. Unfortunately, old-school automation technology isn’t flexible enough to automatically adjust to real-time data or configure for so many variables. And that means IT staffs will have to spend a lot of time manually intervening to get things as close to right as possible if they want to provide their users with choice.

Beyond the POC: Next-Gen Automation for Enterprise-Scale Clouds
IT organizations that use Model T-type automation tools will find that, although they might be able to build successful proof-of-concept clouds, everything will fall apart when they try to scale beyond hundreds of virtual machines.

In order for the cloud to truly take off and deliver on its promises, there needs to be a complete evolution of the automation toolset. It needs to change from a rigid solution that requires more human touch than most IT organizations had planned, into a solution that is context- and user-aware, and can dynamically personalize resources at mass scale with minimal IT intervention.

To handle the scale that enterprises want to achieve, these next-generation automation solutions must:

• Enable IT to define a process that is abstract enough that it can handle different inputs
• Capture business rules and processes at a higher, less literal, level
• Be aware of the user who is requesting resources and the context of each request – for instance, the type of user/job, the user’s location, how the VM will be used
• Synthesize and adapt to real-time factors such as resource availability and changes in context by running discrete workflows that enrich themselves with data on-the-fly, throughout the process
• Retain metadata for the life of those resources and beyond so that it is available to quickly re-create the resources if required

If your solution has these capabilities, you should be able to automate about 90 percent of what your enterprise-wide users require. Not only is that an order of magnitude increase over what earlier products enabled, it’s as much automation as you would want. There will always be the need for some customization that requires the hands-on expertise of your IT administrators. And now, because your staff is largely freed from time-consuming configuration and provisioning drudgeries, they are readily available to handle the edge cases that require more touch.

If you are serious about wanting to adopt cloud computing enterprise-wide and wanting it to be a game-changer for your organization, make sure that your automation solution is designed for scale. Make sure that it doesn’t keep you stuck in the cookie-cutter days of yesteryear. You need mass customization automation capabilities in order to drive the flexibility and efficiencies that today’s dynamic market demands.