Future Facilities Improves Data Center ‘Digital Twin’ Capabilities
Latest 6SigmaDCX release improves power modelling user experience, makes it easier to use simulation data to train deep learning models
January 5, 2019
The team designing your data center is usually different from the team operating it. But their goal is the same: a functional and efficient data center.
In the latest version of its 6SigmaDCX data center design and operations software, Future Facilities aimed to close the gap further by improving the power modelling user experience and DCIM software integration to help customers get a better “digital twin” of their physical data center software.
“A digital twin is a 3D virtual model of the physical data center,” Akhil Docca, Future Facilities director of marketing, explained. “It has all the necessary details, including the IT, the racks, the cooling, and now the power distribution.”
It can be used by both designers, to get the best possible design for their performance goals, and by operations, to prepare for future capacity demands.
Better Power Modelling
6SigmaDCX has been able to model power in several previous releases (as a natural complement to modelling cooling), but the interface could be off-putting, Docca said. The revamped power module should make it easier.
“When they look at a panel for a PDU, it looks like what they're used to,” Docca said in an interview with Data Center Knowledge. “When they connect up items in a logical way, it looks like something they’re used to doing in a tool like AutoCAD.”
That can help with understanding the impact of potential power and cooling failures. “You can map out your cooling and see what happens if cooling fails, and you can also see what happens if you fail a PDU or UPS.”
Future Facilities’ hyperscale customers, for example, use these features to predict the impact of workload changes by simulating transient power demands and seeing whether their cooling systems can keep up, he said.
The new power modelling features also come to 6SigmaAccess, the company’s web-based capacity planning software designed for customers who don’t have DCIM (data center infrastructure management) software. It’s now easier to use data from operations in 6SigmaAccess to engineer future deployments, Docca said.
“The user can drop a [virtual] piece of IT in and quickly connect it up and send it to the engineering model,” he said. “They can take a look at what will happen on the breaker side and the PDU side, now that someone has requested this server be connected up in this rack and in this manner.”
Simulation for Training Deep Learning Models
If you want to move into more advanced prediction and deep learning, the new release (release 13) integrates with tools like MATLAB to send both real-time and simulated data from cooling units, heat exchangers, and IT power to create bigger training data sets for neural networks.
Customers have been asking for the ability to use operational and simulation data to train their models. That way the models could be trained on various possible scenarios without actually experiencing them in the physical facility. “In a real data center you can try maybe ten things and get data from those experiments,” Docca said. “In a simulation you can try a hundred different things.”
Future Facilities may build in native deep learning capabilities in future releases, he suggested. “The most common question we get is, where is the best place to put my equipment. And if you look at all the factors that come into play to answer that question, we’re going to need deep learning built-in.”
New DCIM Integrations
The software already integrates with a range of DCIM software tools from vendors like Schneider Electric and Panduit. The new release adds Digitalor, Flownex, Trellis by Vertiv, and RAMP by Tuangru to the list.
Docca said that the additions were also a response to customer requests and that they go beyond being able to import asset details and live power, temperature, and cooling data. The Trellis support is there to speed up creating the layout for your digital twin, for example.
“One of things that takes time, especially in engineering models, is you have to a build a fairly accurate 3D representation,” Docca explained. “Typically, the starting point is an AutoCAD drawing, and you can start building a room based on that, but if you already have Trellis, we can talk to it and then pull that entire room layout in directly.”
For the next release, Docca said, the plan is for much tighter integration with ticketing systems like ServiceNow and BMC Remedy. “They’re saying, we have this huge wave of requests coming in and we track everything in this ITSM tool.”
That will help to close the loop on using a digital twin for data center operations and management, complementing the ability to predict the future using physics and to bring in live data and live workloads.
About the Author
You May Also Like