Data centers use a lot of power, and while the move to renewable energy helps, other efficiencies must be created to keep these power-hungry behemoths in check. It is estimated that 2% of the carbon emissions in the world come from data centers, and that will only increase as we stream more, save more on the cloud, and demand internet that is faster and faster. However, there is a monster or monsters lurking in nearly every data center – zombie servers.
Jump to Section:
Rather than roaming the world dragging their feet and looking for brains, these zombie servers quietly draw power, making a data center much less power efficient that it could be. It’s not just power either. There is also the hidden cost of the space these servers take up. What are “zombie servers” and what do we do about them?
Even as demand for larger, faster data centers increases, there is a significant push for those data centers to be more efficient. So many data centers use less power as they expand. This is in part due to Power Useage Effectiveness (PUE) standards that determine how well a data center uses the power available to them.
Zombie servers are servers that still use power… but don’t actually do any work. These energy wasters can be difficult to find, especially in large data centers with thousands of servers. Part of the problem is that unlike powering off your laptop, a server is always using energy even when it is “idle.” It’s never truly off. Since the idea is to get equipment to use less energy by doing more work, it is vital that these servers be identified.
The reason quite simply is power usage. Since the server is never truly idle, it is hard to find, because all the servers in a stack may appear to be pulling equal amounts of power, but it can be hard to determine which ones are doing the most work. While AI monitoring helps, it is still not a foolproof solution.
Not to mention the fact that many data centers have not yet implemented that technology. Traditional Data Center Infrastructure Management (DCIM) software can help with power usage and cooling, but in order to achieve newer and stricter efficiency standards, data centers must take additional steps we’ll talk about in a moment.
In fact, as recently as 2015, a study found that 1 in 3 servers in data centers were “zombies”, either comatose or consuming energy and doing little to no work. Immediately, many data centers began to take steps to locate, isolate, and remove these zombie servers. “It’s a matter of technical efficiency,” Jonathan Koomey, a research fellow at Stanford University told Computer World.
Wake Me Up or Just Go-Go
There are two choices when it comes to zombie servers: once located, you can either wake them up and put them to work, or you can move them out of the system. Alternatively, you can move their computing functions to the cloud. The problem is a bit more complex though.
Data centers are often fearful of removing some zombie servers in case they may be needed, or are more mission critical than first thought. This can result in zombies staying online and sucking power longer than necessary. However, as power savings have been proven in other instances, more data centers have become proactive in seeking and taking these servers offline.
What’s the answer? In concept, it is pretty simple. An intelligent DCIM can help by creating certain reports either regularly or on demand. They include information like:
- Charting of trends
- Power capacity trend and analysis
- Power charge back reporting
- Failover testing
- Active power by month and device
The active power by month and device can help you establish baseline power needs and consumption. Then changes can be tracked to determine what servers are underutilized or even not doing any work at all.
The key is a truly holistic approach using all the modern tools at our disposal. The piecemeal approach of shutting down a server here or there will result in only minimal power savings. Optimizing all aspects of on-site infrastructure is essential.
Need help understanding zombie servers? Are you looking for other solutions to free up space in your data center? Or are you in need of remote monitoring and physical network security options? We’re here to help.
Physical Layer Environment Network Security Monitoring and Control
Full visibility, network security and control of your physical layer environment. Monitor your entire hybrid cloud and IT infrastructure from a cloud-based, integrated dashboard:
- Introducing the A150 System
- A150 System Architecture – High-Level Overview
- A150 System Features
- System Controller Hardware and Specifications
- Monitoring Controllers, Probes and Sensors
About the Author
Louis Chompff, Founder & Managing Director, AnD Cable Products
Louis established AnD Cable Products – Intelligently Designed Cable Management in 1989. Prior to this he enjoyed a 20+ year career with a leading global telecommunications company in a variety of senior data management positions. Louis is an enthusiastic inventor who designed, patented and brought to market his innovative Zero U cable management racks and Unitag cabel labels, both of which have become industry-leading network cable management products. AnD Cable Products only offer products that are intelligently designed, increase efficiency, are durable and reliable, re-usable, easy to use or reduce equipment costs. He is the principal author of the Cable Management Blog, where you can find network cable management ideas, server rack cabling techniques and space saving tips, data center trends, latest innovations and more.
Visit https://andcable.com or shop online https://andcable.com/shop/