For all its promise of transforming data center operations to become more resilient, efficient and agile, it’s important to remember that Data Center Infrastructure Management (DCIM) software is primarily an enabling technology. Yes, DCIM solutions leverage automation and real-time telemetry data to provide a holistic view of the data center at the server and rack level. However, most IT staff do not leverage these automation functions at the outset of their DCIM deployment.
As an enabling technology, DCIM software enables staff to reduce or eliminate over-provisioning, to run data centers at higher temperatures thereby saving expenditures related to cooling, and to increase server utilization levels. But if DCIM is so revolutionary, why do so many IT departments become frustrated with its implementation, eventually ripping it out and replacing it, or relegating it to very expensive shelfware?
In my experience, I’ve observed that DCIM deployments tend to fail or succeed based on the relative status of three core issues. Let’s examine each of these issues in their turn.
Most importantly, it’s essential to get executive sponsorship and staff buy-in across the organization to ensure a successful DCIM deployment. Again, as an enabling technology, DCIM is mostly about staff making decisions and taking actions guided by the data a DCIM software solution provides.
Organizational buy-in — starting at the top with the CIO or CFO or both — and C-level communication with and cross-collaboration between the operations and maintenance teams to ensure ongoing support of DCIM, are critical.
Also critical is that any investment in DCIM should be driven by business requirements, and that the selection of a DCIM product or service will meet these requirements over the long-term. Because an organization may not implement all its DCIM software features at once, it’s important to ensure that it’s buying today for the capabilities its business lines will require tomorrow.
Reengineering operational processes
DCIM is not just a deployment or an installation, but an ongoing project. Therefore, if an IT staff doesn’t have policies and operational processes in place to support the software, then it’s likely an organization’s ROI will be nil.
Most enterprises underestimate the amount of process engineering that will be required to maintain their DCIM monitoring systems and asset data accurately. Most frequently, it’s going to be a combination of new processes and reengineered existing ones.
For that reason, it’s critical to have a plan to reengineer operational processes to ensure smooth deployment and maintenance, and to achieve the desired ROI. And reengineering these operational processes requires additional investment in time and other resources, which leads us to the next hurdle to overcome for a successful DCIM deployment.
Staffing, training and budgeting
Before deploying DCIM, you’re going to need to scope and document your staffing requirements, including managing the DCIM software and the data itself. Staff training is typically required.
Even with adequate staff resources, it’s not uncommon for DCIM deployments to lack adequate staffing, once again, because of low levels of staff buy-in at the inception of the project.
It’s also important to budget carefully, and not be surprised if your traditional DCIM software deployment costs and staffing result in two to five times the original project budget. Companies can spend a lot more on services surrounding the software than they expected, including for deployment, the integration of DCIM data with other data, and for ongoing DCIM management and maintenance.
There are of course other hurdles to overcome to ensure a successful DCIM deployment and achieve the desired ROI, including integrating the software with legacy equipment, and the failure to leverage all the software’s features, specifically the IT-facing capabilities, such as cost analysis of the data center and IT utilization levels.
Dave Eastman is Vice President for InCommand DCIM software suite at Serverfarm