By Jonathan Koomey, Author, Cold Cash, Cool Climate: Science-based Advice for Ecological Entrepreneurs
Information technology (IT) has been the catalyst for revolutionary change across modern companies, reducing costs, enabling rapid institutional transformation, and generating new products and services. The great irony is that the IT industry has itself not been much affected by the innovations it enables, with the foundation of enterprise IT being datacenters that embody decades-old assumptions, suboptimal behaviors, and archaic institutional arrangements.
Most companies view IT as a cost center, but the most effective users of these new technologies understand that IT can be a cost-reducing profit center that also improves environmental performance. Reforming IT is imperative for organizations that want to innovate faster and compete more effectively, a fact that modern enterprises ignore at their peril.
The critical factor is senior management attention to restructuring enterprise IT, since addressing these issues is primarily an organizational challenge that is “above the pay grade” of those responsible for datacenter operations. Companies that tackle this challenge will reap rewards in increased revenues, lower costs, faster innovation, and greater business agility. It’s a win-win-win-win, but most upper managers ignore this potential, treating IT as a cost of doing business that they simply must pay.
Many of the biggest names in computing have transformed their IT to generate business value. The best of those companies rationalize their operations to make the full costs and revenues of all choices affecting the delivery of IT services clear to everyone in the company, down to the business unit (and even project) level. These metrics should include costs, revenues, and greenhouse gas emissions per user, per server, per kWh, and per transaction (see here for an example of publicly released metrics of this type).
Once such metrics are instituted, a programmer can know the effects of programming practices on both infrastructure costs and business revenues, which is something never before visible. And once the implications of efficient behaviors are made visible, they can be rewarded, thus incenting a whole systems view that can lead to substantial improvements in the efficiency of computing services up and down the entire stack.
In most enterprise datacenter operations, the key departments and budgets are located in different parts of the organization, thus inhibiting whole systems optimization. One important part of needed reform almost always involves consolidating IT decision-making under one vice president, so that real estate, facilities, IT, and software development all report to the same person. This way, the total costs and benefits of decisions affecting IT operations will be considered by the whole team before making changes. Because the difficult work of tying business key performance metrics to enterprise IT operations has already been done, the company can evaluate results with confidence.
Enterprise IT is often provisioned in a customized way, much as the kitchen at a “sit down” restaurant customizes each patron’s dish. It is almost always beneficial to move from this model, which can lead to weeks or months of project delays, to what one might call “buffet style” IT provisioning. In this model, customized IT requests are limited to a couple of unique server models (instead of the dozens of models that are typically being specified in organizations before equipment ordering is rationalized) in a relatively small number of configurations. The IT organization can keep the standard servers “in stock,” thus speeding up deployments significantly.
An important and needed change is the creation of an “internal cloud,” where many demands for computing services can be met by employees simply typing-in an account number. This reduces the wait time for server resources to seconds or minutes, which has a strongly beneficial effect on business agility. Many applications don’t require planned deployments until they reach significant scale, and smaller-scale experimentation can happen much more rapidly using this approach. It’s also much easier to track resource utilization in virtualized cloud installations, so costs can be more effectively allocated back to those who imposed them.
Another best practice is to move towards more mass production, either using standardized containers or IT and facility deployments that don’t vary much throughout the datacenter. Most enterprise datacenters contain needless internal variations in equipment that contribute complexity and increase costs, but don’t add business value. Hardware is a commodity–differentiation should only be found in the software applications, except in some very limited special cases.
The best existing datacenters deploy state-of-the-art measurement and monitoring tools (often called DCIM, or Data Center Information Management) to track facility conditions in real time. When DCIM is combined with detailed equipment inventory management and predictive modeling (to anticipate cooling and power constraints before IT equipment is deployed or removed), then IT operations are truly operating in a state-of-the-art fashion.
Finally, in this new world, the chief information officer (CIO) moves from being a “keeper of systems” to being a “broker of information services.” CIOs who embrace this new role are the ones whose companies will reap the full benefits of these transformational technologies
Enterprise IT needs to be measured on the profits it generates for the firm. Successful IT deployments will reduce costs and improve environmental performance, but they should also increase computational output and revenues. If you treat IT just as a cost center, you’ll be missing the biggest business opportunity since the advent of electricity (and before that the steam engine). Don’t be left behind. Transform your IT services today!
Jonathan Koomey is a Research Fellow at the Steyer-Taylor Center for Energy Policy and Finance at Stanford University, and is one of the world’s leading experts on datacenter efficiency, costs, and operations. He is the author of Cold Cash, Cool Climate: Science-based Advice for Ecological Entrepreneurs and more than 200 other books, articles, and technical reports. @jgkoomey