Digital Yin and Yang: one part of the data center industry pushes for massive scale compute power and storage facilities and the other is pulling for a smaller, agile, and rapidly deployable data centers with minimal logistics footprint.
Micro and Edge Computing Gaining Traction
For large “traditional” data centers, the global electrical power consumption was found to be 416.2 terawatt-hours in 2016 and 300 terawatt-hours was consumed by the UK in total. Most of the cloud movement has stabilized for the first half of this year but recent U.S. demands are starting to show up in cloud leasing while cloud service providers (CSPs) look to expand into international markets – the drive is due to increased data sovereignty; data privacy and global client demands. The bottleneck has most recently been due to the lack of data center availability and now many are leasing large chunks with 30 percent from cloud users according to JLL Research.
There’s no argument that cloud users are fueling demands for greater capacity, advance facilities with larger spaces and buildings, cutting edge technologies, efficient electrical power and cooling resources, and more in support of the enterprise. However, the other side of the coin has already begun to show us that a greater need exists for the micro data center or edge computing technologies. According to Network World, a shift is occurring away from the “traditional” and centralized cloud data centers and toward a localized, remote data center due to mobile network operators (MNOs) with 5G cellular networks on the horizon. Regardless, there is still a need for a mix – consolidating or distributing, locally and centrally data centers.
What is a Micro Data Center?
According to Gartner’s Hype Cycle for Data Center Power and Cooling Technologies (2016), an MDC is modular or containerized small, shippable data center in a box which may include one rack or less. The MDC contains all the typical IT functions but it is meant to be self-sustaining with respect to heating/cooling, compute power, security, real-time sensing, storage, telecommunications, integration with data center infrastructure management (DCIM), fire suppression, and uninterruptible power supply (UPS). Some products are ruggedized for outdoor deployment with The National Electrical Manufacturers Association (NEMA) ratings and others are man-portable.
Who Uses Micro Data Centers?
To really appreciate how MDCs are being employed, it helps to understand what is driving the need. The major benefit to having an MDC remote located is primarily for very low latency and localized compute power, at the edge. When compared with the latency between the cloud and device, the delays can exceed 100mS but when using the MDC and placed between the cloud and device, latency can be tens of milliseconds. The idea is that by having compute power collocated near the devices, much of the bandwidth required back to the cloud and latencies is negligible.
There are many planned and potential uses for the MDC including driverless cars, IoT sensors, wearables, SMBs (small and medium businesses), Industrial control systems (ICS)/Supervisory control and data acquisition (SCADA), Critical Infrastructure Protection (CIP), Smart Cities, machine-to-machine (M2M), continuity of operations (COOP), streaming videos, life-critical uses and many others. Many in industry have indicated a need for system availability (e.g, 100 percent uptime), tech refresh cycles and hardware standardization without remote management difficulties.
Gartner’s Top 10 I&O Technology Trends include the micro- and edge computing technology category included for remote management of assets with geographically-specific Internet of Things (IoT) requirements. And according to Forbes’ Internet of Things By The Numbers: Market Estimates And Forecasts, there will be 26 billion devices that are Internet connected and $1.9 trillion in global investments.
Another area that has been struggling to become efficient but is nonetheless a bottleneck is in data analytics. Specifically, the difficulties has been in the ETL - extract, transform and load which is a process that must prepare raw data from various sources before it can be processed.
Additionally, the containers market is poised to explode because they are more robust than virtual machines (VM) regarding network function virtualization (NFV) compared to proprietary hardware appliance solutions at the enterprise level. Containers and edge computing solutions are combined to enable data transmissions between the cloud and devices due to extreme scaling. A 451 Research report predicts that the application container market will exceed $2.7 billion with a CAGR of 40 percent by 2020.
Ready to explore Instor’s data center services? Whether your needs are massive or micro, we can help.