Asia's Source for Enterprise Network Knowledge

Sunday, May 26th, 2019

Data Center and Infrastructure

Data Centers in the age of Digital Transformation

1. Are current data centers ready for IoT demands and needs? Are current cabling and wiring sufficient to handle the loads?  

 I think the explosion in the number of connected devices combined with the continued development of data analytics and artificial intelligence, as well as the growth in applications requiring very low latency is leading to a hybrid computing architecture.  This architecture includes the large centralized cloud data centers as well as regional edge and local edge computing sites that enable near-the-end-user processing, storage, and low latency.  Edge computing sites, in effect, take the pressure off centralized data centers and the network itself as IoT demands increase.  5G deployments will further help on this. 

2. How much automation of the data center can we expect IoT to bring in? will it be only patching, monitoring, updating, scheduling, and configuration or will we see the arrival of predictive and preventative maintenance?

We have a very positive, forward-looking view of automation and its role in data centers.  The growth of IoT – connected devices collecting, processing, and sending loads of valuable data – is fundamentally what gives us the possibility to begin to fully understand the complex relationships between the interconnected infrastructure systems. By understanding how systems behave and are driven by all of the variables, we can more fully automate the operations of these system and make that automation the most efficient possible.  Of course, we already have some degree of automation in data centers today. SCADA systems are designed to automatically control the flow of water in a chilled water system, for example, by actuating valves and controlling pumps based on predefined logic and setpoints.  We can do much more, however.  We see the growth of IoT, improvements in data analytics, and the development of machine learning algorithms as the means to making predictive and preventive maintenance a reality in data centers. Schneider Electric is actively working on this today and has begun to provide some of this functionality through our EcoStruxure Platform.    

3. Will smart data centers be the greenest available? How green can IoT make them? 

Yes, generally speaking, data centers that are operated and maintained based on artificial intelligence and insights gleaned from big data analytics are more likely to be more efficient and reliable than those that do not, all else being equal.  It is important to understand, however, that AI and analytics are not “magic pills” that solve all problems. They do not eliminate the need to use the fundamental best practices (eg, air containment, eco-mode UPSs, economizer mode for cooling systems, etc) and energy efficient equipment.  AI and analytics will yield incremental improvements in your efficiency. 

4. How much can we push processing to the sensors in smart buildings? How do we need to rethink security? Is it an added layer or should we move to silo things away from the IoT devices?

We see the deployment of micro data centers as the key means to achieve this. A micro data center is a small self-contained enclosure that includes the IT and all of the supporting infrastructure needed to support it including a rack, power, cooling/ventilation, physical security, environmental monitoring, and management software all in one easy-to-order and deploy solution. Traditionally, distributed IT has not been given the same level of attention and care that you see in larger centralized data centers.  We believe they should be, of course.  Well-designed micro data centers provide security and power redundancy for the critical IT supporting all of the building’s smart sensors and devices. Because these resources are typically so distributed across many sites with few staff to manage it all, remote management becomes a big thing. New management tools are needed to give people easy visibility to the potentially thousands of devices across hundreds or even thousands of sites.  

5. Can current SIEM and DCIM solutions keep pace with the information coming from IoT devices? How should IoT solutions be best integrated into existing solutions? 

Traditional on premise, license-based DCIM systems are fairly limited in their ability to monitor many devices.  They aren’t very scalable.  Our approach at Schneider Electric with our EcoStruxure architecture and platform has been to develop a cloud-based DCIM tool set that we like to call, DMaaS, or Data Center Management as a Service.  Being cloud-based there is no limit to the scalability…to the number of devices that can be managed.  EcoStruxure also includes a concept we refer to as a “data lake”.  It is a safe cloud-based repository for enormous amounts of physical infrastructure device data that is collected by the tools. This data then provides the means to perform data analytic functions and to develop and train machine learning algorithms that then can be applied by the user for useful insights and predictive analytics. So EcoStruxure gives us a secure, scalable, robust cloud architecture, as well as a data lake with massive amounts of normalized data.  This foundation, we believe, is what is needed to address the IoT edge management challenges of lack of staff and multiple sites.    

6. Should we think of these new automation tools like RPA and RDA as the new middleware of the future? What is the best way enterprises should be looking to use them? Are they ready for mainstream deployment or should they wait for more mature solutions? 

Enterprises looking to improve management and operations of their data centers can benefit today from automation, data analytics and even artificial intelligence to some degree at this early point. The industry really needs to continue to develop that foundation – a robust cloud architecture for DCIM and a data lake – something I think Schneider Electric is well down the path on.  So I think data center owners should be considering upgrading their traditional on premise DCIM tool sets they use today and look to moving to one that is structured to take advantage of IoT, Big Data analytics, and artificial intelligence. 


Patrick Donovan; Senior Research Analyst, Office of the CTO at Schneider Electric