Subscribe to the Data Center Knowledge Newsletter
Get analysis and expert insight on the latest in data center business and technology delivered to your inbox daily.
The Robot-Driven Data Center of Tomorrow
Robotics are beginning to be integrated into data center management, creating the potential for a robot-driven, fully automated "lights out" data center environment. Bill Kleyman explores the possibilities.
Bill Kleyman
May 22, 2013
8 Min Read
google-tapelibrary
Tape libraries, like this one at Google, provide an example of the use of robotics to manage data centers. Robotic arms (visible at the end of the aisle) can load and unload tapes. (Photo: Connie Zhou for Google)
There is an evolution happening within the modern data center. Huge data center operators like Google and Amazon are quietly redefining the future of the data center. This includes the integration of robotics to create a lights-out, fully automated data center environment.
Let’s draw some parallels. There's a lot of similarity between the modern warehouse center and a state-of-the-art data center. There is an organized structure, a lot of automation, and the entire floor plan is built to be as efficient as possible. Large organizations like Amazon are already using highly advanced control technologies – which include robotics – to automate and control their warehouses.
So, doesn’t it make sense to logically carry over this technology to the data center?
Robotics in the Data Center
As the reliance on the data center continues to grow, full software and hardware robotics automation is no longer a question of if, but a matter of when, technologists predict. Robotics organizations, like Chicago-based DevLinks LTD are already having conversations and creating initial designs for data center robotics automation.
Scott Jackson, Senior Robotics Programmer at DevLinks, says it’s becoming quite feasible to have a robot fetch a drive, blade or even a chassis and deliver it to a central bay for replacement.
"Simple RFID tags, laser and barcode identifiers can create true data center automation," Jackson explains. "For example, you can tag drives with RFIDs and assign them to be wiped, destroyed and reused as needed." Conveyor systems are able to run in parallel to robotics within the data center environment.
There are already working examples of robotics in the data center. Tape archives seen at Google and high-performance computing data centers use robotic arms to locate and retrieve backup storage tapes. (For an example, see this video of a system in action at the NCAR data center).
What Will Be Different?
What might a robot-driven "lights-out" data center look like? There would be rail-based robotics capable of scaling the entire data center. Here’s an interesting wrinkle: the modern data center would no longer be limited by horizontal expansion space. When using robotics, data centers can literally scale upwards. Utilizing space in the best possible manner is always a challenge for data center providers, so having the ability to scale both horizontally and vertically becomes a huge advantage.
“These robotics can scale the entire rack, which can now be much taller because of these intelligent robots can reach higher," said Jackson. "Once a part is removed, a conveyer at the bottom can move the part to the appropriate floor space. Furthermore, detailed vision technology has progressed a long way as well. Solutions like Cognex are able to allow machines to take pictures of a device, barcode and many other variables to help identify the part’s destination or origin.”
Large organizations that invest heavily in their data center infrastructure are actively exploring robotics solutions to help them better control their data centers. IT shops such as Amazon and Google are looking at ways to create a fully automated, lights out data center. AOL has taken a first in that direction with an unmanned data center facility.
The Cost Equation
As with any technology, costs for custom data center robotics will start high and come down as time progresses and platforms become smarter. Smaller robotics are already becoming less expensive. Manufacturers like FANUC develop large machines; but they also create smaller, more agile robotics. Models like the LR and the Mate M-1iA are paving the way for super-agile, fast, robotics capable of granular part identification and distribution.
Both data center, automation, and robotics technologies have come a very long way over the past decade. From the warehousing perspective, robotics already know where everything is located, how to put things in order and are able to directly interact with the human-created automation scenarios. Because of robotics, something very interesting has happened: Instead of the human going to the warehouse, the warehouse comes to the human.
Soon it will be possible to do this at the data center level.
This would enable entirely new approaches to operations. Your data center will be able to run at a different temperature level, you won’t need any lights, and you can directly integrate your new robotics platform into a modern-day automation and orchestration platform. From a central command center, the human operator can maintain visibility into their data center environment, the robotics infrastructure and the workloads that are being managed. This can all be done without the need of a single person on the data center floor.
fanuc-robotics
Jobs, Challenges and Benefits
Will this technology replace data center administrators? No. However, it may force a gradual evolution within the data center world and a new learning pattern for the future engineer. Human oversight will always remain essential and present. As intelligent as these machines may become, a human will still need to enter in commands and provide some basic maintenance. Data center administrators will need to consider how to stay current and integrate their systems with incoming robotics platforms.
Still, creating such a platform won’t be entirely easy. For one, there really isn’t a blueprint or a schematic for such as design … today. Plus, there are some feasibility challenges to examine as well. Here's a look at the challenges and benefits.
Challenges
Server hardware is not currently designed to be handled by machines. There is a lot of care that goes into handling expensive server and hardware gear. There will certainly have to be some type of adaption from either the server-side or the robotics maker.
Custom hardware would need to be created. To handle robotics, server hardware, racks, and the entire data center itself would have to be redesigned. Not only would new types of environments have to built from scratch, there would have to be some serious throughput into the entire overall design and flow of the data center.
Cabling would have to be specialized. Server blades are one thing. Creating a simple physical switch or latch can allow a machine to insert or a remove a server fairly easily. But what about a backplane – or a switch for that instance? There would have to be careful design considerations around the wiring and cabling process when robotics are in use.
Big upfront cost for automation. Building a robot-driven data center will not be inexpensive. Apart from the physical hardware costs revolving around server gear (some of which may have to be custom), there are costs around the robotics deployment and architecture. Just like the warehouse example earlier, there has to be systematic flow to the data center process. Robotics throws a curveball into the entire process since there must be a very logical run-book in place.
ROI must be understood. A very thorough cost analysis must be done before any such project is considered. For smaller shops, going the robotics and automation route won’t make sense for some time. However, in calculating ROI for organizations like Amazon and Google, the numbers may even out after a shorter period of time.
Benefits
Grow vertically instead of just horizontally. Robotics allows the data center to be extremely efficient with space. After all, robotics will allow us to reach higher and go much further than we’ve been ever able to go. The ability to scale upwards allows data centers to create new designs utilizing floor space much more efficiently.
Pretty much eliminate the need for lighting. Not just lighting, but many other costly data center variables can be eliminated as well. Small comforts which take up space because of the human element no longer need to be included within the data center.
Huge reduction in downtime. Given that there is a structured and fully functional operational flow setup, almost every aspect of data center management can be controlled and calculated. Presently, if a blade breaks, an administrator may swap out that piece of hardware ina matter of minutes. With a robot, every little detail and function is calculated and predictable. There is no guessing – everything can be forecast and controlled.
Machines are able to run at slightly higher temperatures – as can newer server models. The next generations of server technologies are designed to run at higher temperatures, and stay just as efficient. This also holds true for robotics. Within a fully integrated, lights-out data center, environmental variables can be completely shifted towards even more efficiency and hardware control.
Long-term ROI. Although the design and implementation can be costly upfront, the long-term benefits are evident. Administrators will be able to align their software and hardware automation platforms directly with their robotics infrastructure. All would work in unison to create a truly efficient data center machine. With less downtime, more workloads, and easier management, cost savings can be realized within a well-designed data center infrastructure.
There are already some amazing advancements around workflow, cloud and hardware automation. The next step is to integrate robotics into this design and create the data center of tomorrow. So, when will this type of system become a reality?
"I’m not sure of any exact timeline," said Jackson. "I do, however, know that some big data center users in the industry are already testing and discussing these technologies. It would probably take about two years to plan, design and implement something like this. Still, I wouldn’t be surprised if this really started happening within the next five years or so."
About the Author
You May Also Like
