UK & Ireland | English - Change

Does Machine Learning signal the end for humans in the datacentre?

Posted: 28 June 2018 By: Tech Data Categories: Storage

Customers have long sought a simpler datacentre, where costs are continually optimised, service is maximised and the need for human intervention is reduced. Until now this aim has been serviced by incremental advances in technology and processes, but advances in data science and the rise of AI mean things could change rapidly, to the point where humans could soon become unnecessary in the datacentre.

A key change is the scale of data analysis taking place. Solutions such as HPE Nimble storage with InfoSight are analysing data on a global scale, across many customers, maximising available insight.    HPE has analysed over 12,000 Nimble use cases in over 7,500 customer environments, across the entire multi-vendor, multi-site infrastructure stack. The predictive analytics engine InfoSight aggregates and analyses information from up to 70 million data points every day, generating an unrivalled granular view of each infrastructure, and a holistic view of every infrastructure where Nimble is present, learning lessons as it goes.

Together Nimble and InfoSight employ machine learning to develop an understanding of potential issues arising that is beyond the reasonable capability of a human team. In 90% of cases InfoSight is able to detect issues and remedy them before customers even realise a problem exists.

This global data analysis and machine learning activity has uncovered some surprises, for instance humans usually suspect storage  issues are driving the gap between applications and data delivery. However InfoSight has learnt that in fact fewer than 50% of cases are caused by storage; instead 54% of problems arise from poor configuration, interoperability and lack of best practice. Whilst 46% are storage issues arising from related hardware and software and software update issues.

For humans the implications are that the technology is already accurately predicting and fixing potential and actual failures faster than we can. It’s learning about current failures, and what future failures might be. It’s giving them a signature, and rules for resolution, then replicating these into other customer sites faster than a human team ever could. The AI is also driving future software updates, implementing what it’s already learnt.

In practice it will be some time before we are eliminated from the datacentre but organisations are already employing autonomous infrastructure to reduce the size of their support teams, relying on AI to fix the 80% of issues that are simple and common, before they even impact. For the 20% of issues that are complex and unique, customers are employing a lower number of higher skilled people to deliver the right level of support, a trend that will gather pace as the AI technology evolves further.

For more information about AI in the datacentre, watch this HPE video. Find out more about HPE Nimble here.

Back to blogs