A Manifesto for Efficient IT Infrastructure Management

January 9, 2012 3 Comments »
A Manifesto for Efficient IT Infrastructure Management

Data Center Infrastructure Management

Data center infrastructure management (DCIM) is about more than just managing environmental data for the assets inside the live data center space. DCIM must include all aspects of the equipment in the data center, extending to the entire lifecycle of the asset from purchase to end of life. In fact, I would suggest that DCIM is actually a pretty poor term for what should really be ITIM (information technology infrastructure management), as it’s really about managing in terms of the IT, not the data center. Data center managers face more challenges today than ever before, including increasing demand for IT applications and equipment, pressure to avoid downtime and demand from upper management for better operational efficiency. With the majority of companies still using spreadsheets to manage their data centers, it’s no surprise that IT leaders grossly underestimate the impact a poorly managed data center has on the bottom line.

Imagine your staff is ordering equipment for your data center and the equipment is going right to the store room, not just without being deployed, but also without being inventoried. Then it sits there, depreciating in value and collecting dust. This is just one of the horror stories I’ve heard in the field. The company this was happening to ultimately discovered that, in just a six week period, it had accumulated more than $700,000 in depreciation costs for assets that were not being used.

This is where managing the complete lifecycle of assets comes into play. IT managers need to be thinking about planning where each new asset will be used from the moment it has been ordered and well before it has arrived. They need to be able to look at the data center and see real-time, historical and future data related to all aspects of management, including power, space, environmental factors, location, connectivity and ownership. Then they can plan and allocate space for the new asset and ensure it is deployed and correctly verified as ready to enter an operational state as soon as possible. The data center manager needs to document this process so that when a problem occurs, he or she can use the same information to troubleshoot the issue and resolve it as quickly as possible. Even virtualization is involved here. As severs are virtualized, the same data is needed to decommission the right servers at the right time.

One of the most common, and costly, issues facing IT managers is data center downtime. Studies show that it costs companies more than $5,000 per minute, with the average incident lasting 90 minutes, resulting in an average cost per incident of approximately $505,000. [1] But all of this can be minimized if the IT infrastructure is properly managed.

The key point to remember is that managing is different than documenting. Documenting is gathering the data, and if the vehicle for this documentation is spreadsheets, that is often where the process stops (if it happens at all). In my experience, even with great spreadsheet records, each planned change to the data center is accompanied by a comparison of the real infrastructure and the infrastructure records, just to be “sure”—that is hardly management! Management is when you can look at the data center from a holistic point of view and say things like, “Let’s look at what has been deployed recently, what capacity is available, what planned deployments are coming up and what are the capacity constraints.”

Case in point: one company I spoke with had a smaller data center of about 80 racks. It was keeping records in an MS Access database and was convinced the records were accurate. This company was more diligent than most, running monthly checks to ensure the data was correct. These checks, however, were being performed on the basis of paper printouts of the data, which in theory was updated into the Access database from hand-written updates made to the records. When a proper full audit was performed, the team found 63 servers and switches that were not recorded anywhere, further illustrating that home-grown or spreadsheet management is not suitable for enterprise-level infrastructure management, nor can it yield maximum efficiency.

From the C-level executive down to the IT manager, data center efficiency is top of mind for everyone these days, as energy, people and space efficiency are business imperatives. It’s no secret that data centers consume a lot of power, with estimates putting it at about two percent of global energy consumption. [2] IT managers can take several steps to improve energy efficiency, starting with an assessment of the current data center conditions. Are there unused servers that could/should be shut down? What are the best candidates for virtualization? What, really, are the least efficient pieces of equipment in the data center? Without truly managing the infrastructure, determining problem areas can be next to possible.

I was recently talking with an IT manager who was having issues with power overloads to the point where electrical breakers were tripping because of current overloads. The data center was an older design and did not have end-of-row power panels; rather, it used power fed from distribution boards. His first step to fix the issue was to commission an auditing and tracking exercise of each power circuit, using spreadsheets to document the information. After spending approximately $20,000 for external electricians, and after moving equipment (at the cost of additional internal staff time and some service downtime), the problem was resolved. But within three months the problem started again, as new equipment had been deployed, and the spreadsheets, having not been maintained, did not reflect the current deployed equipment status, leading to another costly audit.

The Manifesto

Steps must be taken to manage the IT infrastructure (whether you call it DCIM, ITIM or something else) more efficiently. Start by deciding what data is important to managing the data center efficiently. There is no single list and no one-size-fits-all answer. It all depends on the age, size, scale and equipment in each unique environment. Then, pick a management system that will truly consolidate all the data currently managed in spreadsheets, as well as the data that should be managed but isn’t because it’s too difficult. Strongly consider whether the solution requires hardware to be purchased and whether this will add to your problems or lock you into one system. Next, ensure that real, positive processes are in place to keep the data up to date; no tool is of any use if keeping it accurate is too much trouble. Some systems make this easier than others. This step is critical, as no tool by itself—no matter how fancy and regardless of what the vendor says—is going to solve problems by itself. The next step, using the data for planning and further change, requires trust in the data. Avoid falling into the trap of using the tool and still doing a physical audit each time change is required. All of this can be achieved, but it’s not easy, and there are no magic bullets. Do not believe a vendor that says its system will solve all your problems without process changes; it won’t!

Properly managing your infrastructure is critical to avoiding downtime as well as meeting operational efficiency demands. If you are cutting your data center short on proper management to conserve money today, it is almost certainly costing you more in the long term, both in resources and downtime. Whatever the future plans for your data center, whether it is a move to virtualization, trying to become more green or adopting a cloud model , having up-to-date, central records of all aspects of your assets, the facility, connectivity and historical data (not in spreadsheets!) will help you make those transitions more smoothly. Although no two enterprises are ever the same in what they want to manage, the principle that change cannot be managed effectively without accurate consolidated data and a truly effective infrastructure management system applies to everyone.

References

[1, 2] Ponemon Institute, Calculating the Cost of Data Center Outages, 2011.

About the Author

Paul Goodison is CEO of Cormant.

infrastructure management

 

About Jeff Clark

Jeff Clark is editor for the Data Center Journal. He holds a bachelor’s degree in physics from the University of Richmond, as well as master’s and doctorate degrees in electrical engineering from Virginia Tech. An author and aspiring renaissance man, his interests range from quantum mechanics and processor technology to drawing and philosophy.

3 Comments

Add Comment Register



Leave a Reply