The place for AI in DCIM

For film aficionados, the concept of Artificial Intelligence (AI) is as novel as seeing popcorn being sold in a cinema foyer.

From HAL 9000 – the chief protagonist in the 1968 classic 2001: A Space Odyssey – to the Avengers’ J.A.R.V.I.S, machine vision and learning have been a mainstay of movies for decades, with many a plot line exploring the possibilities – and potential perils – of smart supercomputers and sentient robots.

In sharp contrast, when it comes to exploiting such technologies, the real-world – of which data centers are an increasingly prominent feature – has long lagged behind the box office curve and even today general AI is generally viewed as an aspiration rather than actuality.

The buffering in breakthroughs is, however, fast being addressed, with another sector of the entertainment business – the computer games industry – spearheading a surge in development. 

Indeed, AI is already more prevalent in our everyday lives than some might appreciate. It is busy behind the scenes, determining social media feeds, suggesting streaming service subscribers boxsets to binge on and writing computer code for popular apps. 

AI is also providing the brains behind the mechanical brawn of assembly line robots in factories; delivering safer roads through its application in driver-assistance technology; detecting and preventing cyber security threats; and saving lives by intercepting missiles in war zones. 

This summer it even successfully navigated a crew-less ship – designed to recreate the Mayflower’s historic journey across the Atlantic 400 years ago – on a 2,700-mile ocean crossing. 

And if the technology can conquer the high seas, it stands to reason that AI can also excel in the ‘Cloud’ – or more accurately the data centers enabling the infrastructure on which so many businesses and organisations rely.

The rapid expansion of these facilities – which encompass public cloud, co-location, enterprise and edge sites – to meet society’s insatiable demand for connectivity and the multifaceted nature of their environments make them ideal candidates for computer-assisted control.

Allowing a machine to do the computational heavy lifting in a data center, generally home to multi-generational systems and software and hardware from multiple vendors, could extract clarity from the complex.

Given the volume and wide spectrum of sources of operational information available for analysis across a facility’s grey and white space, the ability to extrapolate and predict unknown data would serve as a reliable roadmap to optimisation.

Integrated correctly, AI could be used to identify trends and risks and suggest solutions that guide operators to unprecedented levels of effectiveness and efficiency in the utilisation of space, power and cooling.

It is for such reasons that AI capabilities sit at the cornerstone of UIIM (Universal Intelligent Infrastructure Management) solutions, which are being developed to ‘out-think’ and supersede the intelligence of traditional DCIM tools.

The tech-wary reading this may be quick to counter that the means to attain such a nirvana already exist and – as highly-trained human experts – are literally alive and well. However, the mental agility of specialist technicians, engineers and operational strategists is not the issue; their availability and inability to be omnipresent is.

Round-the-clock vigilance is neither feasible or a cost-effective answer and therefore efforts to optimise are not exhaustive and generally considered from a limited perspective. Yes, common trends can be manually identified, and action taken accordingly, but the impact of any resource decisions are rarely completely understood.

Delivering operations in a multi-vendor environment that is home to multi-generational systems and in which conflicting departmental agendas can come into play can result in changes intended to enhance having an adverse effect – leading to risk and economic and environmental inefficiencies.

As well as helping to preside over the present, the embrace of machine learning would minimise data gaps and guesswork in respect of future proofing.

Having interrogated its surroundings to develop a universal understanding, AI could perform predictive planning – forecasting the lifecycle of a data center’s devices, mapping the implications of any retraction or expansion of services and proactively identifying opportunities to enhance performance and reliability or optimise costs.

Such models should exploit historic data to do so; identifying patterns from the past and measuring the effectiveness of previous interventions to proactively inform and refine resolutions.

Such a data center “megamind” may not yet exist and there remain substantial challenges to creating one, chief among them issues over data reliability and the absence of a platform and vendor agnostic language for it to communicate with, but leaders in the sector should be mindful of the innovations currently available if they wish to be best placed to exploit the advancements of tomorrow.

AI-driven automation is already a reality and a feature of some Data Center Infrastructure Management (DCIM) tools. 

Unfortunately, a succession of mis-sold, mis-marketed or unfit for purpose platforms has bred a mistrust of innovative solutions and reticence to reinvest in DCIM developments, but those who dismiss refinements to the concept as the equivalent of purchasing a modern-day equivalent of the “emperor’s new clothes” are making a monumental mistake.  

There are ample rewards on offer to technologists willing to embrace change. AI-based automation can be used to enrich data and bridge gaps in device intelligence.

XpedITe, RiT Tech’s next-generation DCIM solution, for example, uses advanced algorithms to determine provisioning for new assets at speed; assess the risks and rewards to come up with optimal decisions; and produces automated work orders that signpost engineers directly to their “to-do list”. 

The generation after next of data center AI will undoubtedly go much further so selecting a provider that has an eye fixed firmly on the technological horizon will ensure any financial outlay made today does not end in obsoletion. 

That is why selecting a vendor with a clear innovation roadmap, that is abreast of the advantages of automation and able to not only install a solution but oversee its integration, customisation and implementation is crucial for long-term business results.

RiT Tech is heavily invested in a number of major research and development projects in the machine learning field but, contrary to cinematic clichés of rogue robots and AI programmes running amok in an attempt to seize control of civilisation, our endeavours to improve machine intelligence is not a precursor to any form of Doomsday scenario.