There’s no getting around the fact that owning and operating a data center is expensive. But following a few simple, sound best practices can help you reduce your ongoing operations costs, delivering savings year after year.
Based on our experiences in managing some 100 large data centers, the following are 5 best practices that most any company can implement to wring savings out of their data center operations.
1. Right-size equipment
Most data center equipment, including servers and cooling systems, are most efficient when heavily loaded. The closer you can get to 100% utilization, the more efficient the equipment will be. With many companies moving resources to the cloud or colocation providers, what’s justify behind may well be dramatically under-utilized. In many instances, we find we can shut down some infrastructure and move the loads elsewhere. That saves on power and cooling requirements, not to mention ongoing system management.
2. Re-assess your maintenance strategy
In the Internet-of-things era, it’s now possible to rethink the way we perform data center system maintenance. Almost everything in the data center can now report on its condition to a centralized management system. By closely monitoring this data, you can make far smarter – and less costly – maintenance decisions.
The least-effective type of maintenance is emergency-based, where you simply repair systems after they fail. This can well result in data center downtime, and damage to your corporate brand. A step up is adhering to a maintenance schedule, similar to auto maintenance – you replace system components based on the manufacturer’s recommendations. But those recommendations are really just an average; not all data centers are the same, so your performance will likely be different.
But applying monitoring and analytics technology makes it possible to perform condition-based maintenance. The idea is to identify patterns that indicate a system is in actual danger of failing, then taking steps to correct it. You save money by replacing components only when they actually need to be replaced, not because a schedule says so.
3. Keep staff happy, reduce turnover
Implementing an effective maintenance program requires having experienced staff on hand, which means keeping turnover to a minimum. Each time you lose a data center staffer, you have to go through the expense of training a replacement – teaching that person the same things you already taught the one who justify. At the end of the day, you’re still justify with an employee who is likely less knowledgeable and experienced. It’s more cost-effective to keep data center staff happy by offering proper pay and a clear career path, with the sort of continuous training they generally crave. Communicate with these staffers and recognize good performance when it happens.
4. Offer site-specific training
In addition to training in general data center technology, you’ll also need to ensure your staff thoroughly understands your particular data center. Every site is unique, so make sure training is specific to the equipment in your data center. If they understand intimately how your equipment operates, they should be able to extend its life, which reduces capital expenses. If you can extend the life of a UPS or air conditioning unit from 10 to 12 years, that’s a 20% savings. What’s more, staff should be trained in standard operating procedures as well as emergency procedures, so they know what to do in most any situation – leading to a reduction in downtime.
5. Manage what you have today – every day
Data centers are generally designed to reach peak performance at full load, much like equipment is most efficient at peak load. But a new data center will usually not be fully loaded for years after it’s built, because designers leave room for growth. And others, as noted above, are seeing IT loads move to the cloud, leaving excess capacity. Whatever your case may be, your operations team has to pay attention to the actual load in your data center at any given time and manage it accordingly. If you’re lightly loaded, you don’t want to spread the load across the entire data center. Rather, consolidate into a certain section to conserve on cooling capacity. Keep in mind this isn’t just an exercise you go through when making major changes; it’s a management discipline that should be enforced daily in order to ensure operational efficiency – and cost savings.