175 zetabytes. It’s the projected size of the global datasphere by 2025 and a rallying cry for a data center industry determined to keep pace.

Data volumes are exploding. The Internet of Things (IoT) is ramping up machine-generated data at the same time consumer-driven transaction, media, and other information continues to grow. As data lakes become data oceans, what will it take for India’s data center industry to catalogue, store, and analyze the vast quantities of information the nation’s organizations and individuals are collecting? 

Digital transformation has brought technology to the center of the economy, and top-performing data center facilities will be essential in achieving recovery and expansion in the coming months. But rapid shifts in SMB and enterprise customer requirements are redefining “world-class” infrastructure around four key watchwords.

Decentralized

Expansive volumes of data challenge enterprise and telecom/cable/satellite networks. Much like exponential growth in personal vehicles can quickly overwhelm roadways, a similar trajectory for data will clog existing and next-generation networks. No longer can every byte of useful information be sent to a centralized facility for processing.

The answer, in many instances, is to move to compute and, to some degree, storage closer to where the data is created. This imperative has driven attention to edge computing solutions and local and regional data center development. 

Providers in various locations are, or soon will be, intensively planning and partnering with government officials to site data center corridors to support the full spectrum of data center needs. This will be a cooperative project to tap into or build the redundant-provider, high-speed fiber internet access, reliable water service, and plentiful electricity beyond the major metropolises. Data centers will proliferate in second- and later third-tier population centers and micro-data centers will increasingly colonize the edge.

Rather than offering space in public and private clouds or colocation facilities operated solely in a few big cities, data center providers will soon be selling a wider variety of solutions aligned with the customer footprint across the country. 

Powerful

Rising data volumes challenge existing storage capacity, but perhaps the greatest test for the data center industry will be in meeting processing demand. 

After all, raw data is of little value. The interest in compiling more information is no mere hoarding instinct triggered by plummeting HDD costs, but rather a response to increasingly powerful big data analytics, complex modelling, and artificial intelligence capabilities coming online. It’s all about what we can do with and learn from the data.

The consequence is a massive rise in high-density computing, which puts its own strain on data centers. Heat dispersal, for example, is a major concern, and many data center providers are building and retrofitting for facility-wide liquid cooling, or barring that option, integrating liquid cooling at the rack level. 

In the initial stages, supercomputing will be primarily concentrated in centralized and often hyperscale facilities. Data will undergo initial processing at edge and regional data centers, and more refined information will be transferred to traditional data centers for long-term storage and backup. Siting analytics and AI solutions in these locations will limit the need to send expansive data sets across the network backbone and engage the most powerful processors available.

Over time, however, the need for low-latency decision-making—as for autonomous vehicles, smart utilities, automated factories, and so on—will push computing demand toward the edge as well. The strategies used to manage high-density, power-hungry compute within existing data centers will need to be adapted to smaller and eventually extremely remote operations. 

Automated

The trend toward data center automation was already in evidence before the coronavirus pandemic erupted. The current global health emergency, however, can only accelerate providers’ transitions toward self-healing, “lights off” data center environments, which won’t require regular visits from us vulnerable human minders.

But of course, automation is more than a survival strategy against nationwide lockdowns. Emerging technologies promise to ease data center administrators’ alert fatigue and enable granular, realtime tuning to achieve new heights in efficiency. With no end in sight to data center staffing shortages, multiplying the impact of expert personnel with AI will also be a necessary coping strategy to stretch available talent as far as possible.

Fortunately, many of the building blocks for the automation revolution are in place, with more being trialed by hyperscale providers with the budget to push the technology envelope. Advanced monitoring systems have obviated the need for many field technician check-ins at remote facilities. Such solutions will play a key role as edge-based micro-data centers proliferate, delivering a cost-effective, staffing-efficient means to maximize uptime beyond centralized facilities.

Moreover, remote diagnostics and repair are tracking with monitoring advances, improving data center administrators’ ability to manage the technology stack from anywhere. The missing link so far has been “hands on” changes, such as drive swaps or server replacements, but recent robotics solutions, like this one, indicate that the fully machine-operated data center is closer than we may think.

Optimized

Today, data center providers are measured in large part on their reliability numbers. Customers expectations have creeped into six-nines (99.9999% uptime) territory. As the changes outlined above continue to unfold, data center providers will be held to similarly high standards as their facilities multiply across the physical landscape.

The response must be a fully optimized infrastructure on the provider side ready to pair dynamically with the increasingly hybrid enterprise with its public and private clouds, on-premises and colocation needs, and SaaS, PaaS, and IaaS solution. It’s a tall order, but one the data center industry and its own vendor networks are preparing to meet.

Disclaimer: The views expressed in the article above are those of the authors' and do not necessarily represent or reflect the views of this publishing house