Enterprise data architectures will have to get a lot smarter in order to survive in the emerging digital economy. How smart? Smart enough to learn by themselves, apparently.
Machine learning is starting to gather steam as the next big thing in the enterprise, according to a number of sources. The idea is that rather than human operators manipulating resources and architectures from above, the resources themselves could monitor their own surroundings and make whatever corrections are necessary to improve performance. This is part and parcel of the broader Big Data/Internet of Things movement that leverages voluminous data streams and advanced analytics to gain deeper understandings of complex systems, but it also incorporates advances in neural networks and other technologies to push automation and self-learning to new levels.https://o1.qnsr.com/log/p.gif?;n=203;c=204663295;s=11915;x=7936;f=201904081034270;u=j;z=TIMESTAMP;a=20410779;e=iAlready, a handful of software developers are starting to redefine productivity applications and other services around machine learning, says TechCrunch’s Ajay Agarwal. The interesting thing, though, is that the pioneers in this field are not SAP, Oracle and the like who have spent years developing database and workflow expertise, but users like Facebook, Google and Twitter that have built up in-house proficiencies in advanced systems analytics. Much of this knowledge has already been spun off into a range of start-ups like Lattice, Captora and Adaptive Planning that are targeting sales, finance and other key functions.
Some of these start-ups are turning their attention to how machine learning can improve the operation of traditional enterprise platforms. SIOS Technology just released its SIOS iQ solution that uses machine learning analytics to oversee the dynamic environments that arise within virtual infrastructure, namely VMware. As the company’s COO, Jerry Melnick explains to Enterprise Tech how the system provides a new level of deep learning that enables the overall environment to educate itself on normal operations and the impact that major or even minor changes will have on the smooth flow of information. In this way, SIOS iQ can drill down into the root causes of problems and then correct them rather than simply bombard technicians with multiple alerts that something is wrong somewhere.
Much of the functionality behind these advanced capabilities will reside on the machines themselves, or more accurately, on the silicon that powers the machines. Nvidia is working on new software that supports GPU-based deep learning and neural networking that will hopefully lead to full-scale artificial intelligence. The company recently released version 2 of its Digits software with a GUI that is intended to expand AI programming beyond the limited pool of academics and researchers who specialize in the field to the broader data community, says IDG’s Joab Jackson. At the same time, the software now incorporates multi-processor functionality that allows each device to learn from one another.
Intel is also said to be working toward machine learning, but from a different angle. The company’s interest in field programmable gate arrays (FPGAs), punctuated by its recent $16.7 billion acquisition of Altera, suggests a desire for highly malleable processing solutions that can be updated and reconfigured on the fly, says The Platform’s Timothy Prickett Morgan. From there, it’s only a small step to having one set of processors fueled by Big Data and advanced analytics to reconfigure others in order to learn about and respond to changing data environments.
The idea of a self-learning, self-managing enterprise infrastructure may be unnerving to some, but the fact is that the rules governing this kind of functionality will still be set by humans, and it will undoubtedly follow the same development and deployment process as past technologies to accurately determine what works and what doesn’t before it becomes broadly available.
Ultimately, however, machine learning and the neural networks and other advancements that support it offer the possibility that before too long, enterprise data infrastructure will become a full-fledged partner in the business process, not merely a tool to be leveraged.
Arthur Cole writes about infrastructure for IT Business Edge. Cole has been covering the high-tech media and computing industries for more than 20 years, having served as editor of TV Technology, Video Technology News, Internet News and Multimedia Weekly. His contributions have appeared in Communications Today and Enterprise Networking Planet and as web content for numerous high-tech clients like TwinStrata, Carpathia and NetMagic.