When does it make sense to upgrade to more efficient servers? Most data centers operate on a 3 to 5 year tech refresh cycle. Is this really the best way to decide when to refresh old equipment? By continuously monitoring the cost to run older equipment, you can determine when you have hit the break-even point with your existing servers. Join Viridity co-founder and CTO, Mike Rowan, as he reviews best practices for technology refresh.
Who are we? Viridity Software is the leader in Data Center Infrastructure Management Software. Our software connects equipment utilization and business value to power consumption in the data center.Customers are finding time to value in hours, not months.And, because we are a software solution, there is no additional hardware or agents required.
Machine Age is not enough.- Many companies actually save significantly more money in energy reduction than they spend in new equipment purchases. Sometimes the results are staggering. - Last year, for instance, Intel reported that its equipment replacement plan had saved the company $250 million. These kinds of results can occur only when the company is committed to a strong energy management strategy.
System configuration and usage are clearly the determining factors in power consumption, and their financial impact requires some scrutiny. Understanding the average and peak utilization ranges of servers and racks is critical in determining which systems are the most expensive to run. Their cost may have little to do with their age.
Data centers tend to maintain a number of servers that no longer do any useful work for the organization. That number can be quite large: typically 10 – 15%, but often as high as 30%.Orphaned hardware is generally older and has no energy-saving mode when idle. It burns power at a maximum level, even when it does nothing. New equipment, on the other hand, can cut its power demand by as much as 25 – 33% when idle. The resulting power savings can be substantial. Converting large servers to virtual machines can yield up to $4,000 per machine annually.
Other best practices…
Knowing your power curve is key. Newer are dynamic, adjusting their power consumption according to the ebb and flow of their activity level. Operating costs accurately reflect the amount of work being accomplished.System-configuration details are necessary at this point. The CPU is a main indicator of a system’s power curve. Which CPUs are involved?How many cores exist? How fast are the chips? What about memory? What drives are installed? What chassis and fan setup is used? Keeping track of all these details can be daunting.Even a single-vendor system can have so many variables. Most systems have upwards of 50 different CPU options alone.That’s where we come in.
- The right equipment-replacement project may not only reduce the expense of air conditioning, it will also limit heat-related outages. - If a cooling plant is already running at full capacity, it can put off the need for a costly plant upgrade.
Data centers often adopt an all-or-nothing approach about application sources, when their best option might be to house less critical applications outside the data center, and focus on the most critical operations.
--- Improvements in resiliency, which are really more significant than just cost; they are necessary for the health of the business. Equipment-support costs, including new-equipment costs spread over three years as well the projected support costs of the old equipment. Estimated utility savings, both in terms of cost and infrastructure capacity. Estimated protected-power savings. Cooling-capacity savings, again, both in terms of cost and infrastructure capacity. Floor-space capacity, including consolidation benefits, rack-unit calculations, and the advantages of virtual technology and cloud computing.
Key Points:EnergyCenter is Viridity’s Core PlatformIt provides end-to-end capabilities to understand actual power consumption and utilization at the IT asset levelIt further provides actionable reporting literally within a few hours of beginning deploymentPrevious Notes:Frictionless DeploymentDeploys in minutesNo hardware sensorsImmediate ResultsActionable reports within hours of first DiscoveryInsights across IT and Facilities infrastructureDirect ROI