Machine Learning enhances data center performance

Machine Learning enhances data center performance

 

Machine Learning is arguably the most powerful tool we have seen so far for unlocking the potential of Big Data, and one of the sectors that stands to benefit most from it is the data center industry.

Given that most companies are highly reliant on the performance of their data centers to ensure services can be delivered and productivity maintained, there is a constant, acute pressure on facilities managers to predict, manage and react to any change that could adversely affect operational availability and performance.

As a result, data centers are starting to realise the potential of Machine Learning technologies to automate and improve their ability to keep these facilities performing optimaly. Furthermore, even though many companies still use rudimentary tools to track and oversee data center performance, it is becoming increasingly difficult for humans to manually analyse data directly from DCIM systems and verify if the data collected by the vast sensor networks monitoring these facilities is accurate.

Only last month, the Royal Society published a report that looked at the power and promise of Machine Learning. Its focus was on how this area of computing will reshape the UK economy and people’s lives. The report also asked questions about who will be affected, how the benefits will be distributed and where the opportunities for growth lie.

The benefits for data centers have already been proven and the knock-on effects are many and various. Google recently explained that it is using its DeepMind Machine Learning technology to manage power consumption at its data centers by dynamically tuning their performance to reduce their operational energy consumption. In one of Facebook's data centers meanwhile, its Big Sur servers are training Machine Learning systems to ‘read’ images and videos to the blind and display over two million translated stories every day.

In the same way, Machine Learning has a big part to play in further enhancing analytics tools for data centers. The ability to generate validation models for performance analysis, investment analysis and even for assessing the suitability of a location for a new data center offers immense advantages such as reducing costs and improving operating performance.

Our own analytics platform identifies anomalies and can help track down the causes behind a symptom based on comparison to a calibrated predictive model as well as aggregated data ‘knowledge’ collected over years from over 350 data centers around the world. In this way we’re able to provide our customers with powerful and actionable insight on their operations and systems so they can maintain peak performance at all times and increase ROI.

CBRE | Romonet’s Analytics Platform: New Machine Learning Capabilities to be Launched

CBRE | Romonet's Analytics Platform New Machine Learning Capabilities to be Launched

machine-learning-romonet-bannerCBRE | Romonet today announced it is filing for a number of new patents for the next phase of its data center intelligence platform, utilizing the applications of Machine Learning. The company is known as the leader in data center analytics and this development enhances the value of CBRE | Romonet's already patented solution.

"We have been working on advanced data handling and Machine Learning algorithms for over a year, focusing predominantly on enhancing our solution to learn and become as proficient as our human data scientists are today at identifying anomalies, and tracking down the cause behind the symptom. This capability provides powerful operational and business insight into data center systems and component level performance," said Liam Newcombe, CBRE | Romonet's co-founder and CTO.

Having modeled, collected data and analyzed hundreds of data centers in the past eight years, CBRE | Romonet's platform has an incredibly detailed and expansive data archive on how facilities of every size perform under different climate, environmental, energy, IT and commercial factors.

While the use of Machine Learning applications is not new, CBRE | Romonet's platform is the first to combine metered data, Machine Learning, simulation and predictive analytics.

"In our case, teaching the machine is much faster as we feed it pre-cleansed and calibrated data to recognize and learn patterns, incorporate additional data from outside sources, and teach the software to suggest causes and recommended actions from previously learned results," continued Newcombe.

Like Google, Amazon, Cisco and Netflix, who already use Machine Learning to personalize services and business intelligence, CBRE | Romonet's industry-leading platform is revolutionizing the booming global data center market.

With CBRE | Romonet, Hyperscale and Multi-Tenant Data Center (MTDC) operators are improving the services they provide to their customers while strengthening financial management through investment, cost and margin analysis. Enterprise data center owners, whose facilities, while core to service delivery, are also a drain on profitability, are rationalizing their investments, accurately planning a hybrid (owned, colocated and cloud) strategy years into the future, and improving their ability to make socially responsible decisions that impact the environment, shareholders and employees.