Containers, automation and AI shape 2020 IT trends

Stuart Burns

Docker and Kubernetes will become even more popular. Companies are beginning to realize how useful it is to have version-controlled, lightweight containers that can be spun up in seconds. The technology is proven beyond a doubt. It is more a case of how companies are going to take the technology from evaluation to production.

The Docker and Kubernetes stacks are themselves an open source technology. Large companies with the appropriate skills and resources are building their own stacks. This presents both a risk and an opportunity to established hypervisor vendors who are still working out their transitions to ultra-lightweight instances and value-add propositions. While these open source stacks themselves might be free, the management can be complex. Tooling for operational use presents a value-add that the open source, free-to-consume stack does not.

Brian Kirsch

IT operations has been under continual change for years. First it was virtualization, then the cloud and now extensive automation. The good thing for ops is that many of these technologies are no longer bleeding edge -- they all have some level of history now.

First it was virtualization, then the cloud and now extensive automation.

2020 is going to be a year for ops to not just play catch-up, but to gain a working knowledge of tools and techniques for automation and the cloud. This doesn't mean the next new thing won't require some catching up, but the base that ops has to work with will be more solid than in previous years. It will be a year of updates and upgrades, staff trainings and tweaks to make it truly an environment that they have been looking for. It will be up to management to help fund these upgrades and staff trainings -- and that should not be delayed. Peaceful moments in IT are rare and you have to pounce on them because change is right around the corner.

Clive Longbottom

Enterprises will need to answer the uncertainty around plans for data center changes. For example, should they plan to update on-premises infrastructure, extend an owned data center facility or use colocation resources alongside an owned facility? Alternatively, should they run new cloud resources alongside on-premises ones, or plan a phased move to the cloud entirely? Each option has its pros and cons -- and the wrong choice could be disastrous.

Second, enterprises will need to monitor and measure what is happening across an increasingly complex hybrid IT environment and then react to those events. This will require machine or deep learning, AI and admins taking a step back to allow automation to take control in a much greater way. Manual approaches just won't be able to deal with things in real- or near-real time: Fire-fighting will not be a valid way forward as the platform expands to include more areas, particularly as IoT really takes hold.

Another 2020 IT trend will be the continuing need for techies to better understand business goals, be the Babel Fish to translate those goals and identify the best technical means to meet them. This has nothing to do with technical elegance: It is about what meets the business variables of optimum price, optimum risk and optimum overall value. Chasing technical elegance is fast becoming the best way to find that you need to polish up your resume.

Will Kelly

DevOps and cloud management practices will usurp ITIL: The true cost of the eight-year slumber between ITIL 3 and 4 will become apparent in 2020. We'll see ITIL increasingly in a fight for relevance against DevOps and cloud management in commercial and public sector enterprises. While some ITIL principles will survive, enterprises will look to DevOps and cloud management practices to meet the delivery velocity their customers and corporate boards demand.

A market shakeout among startup DevOps tool vendors will hit high gear in 2020. The market shakeout will start for a few reasons. First, there is the natural mergers-and-acquisitions rhythm of the IT world. Some smaller, more innovative tool vendors will get snatched up by larger players, including private equity firms. Second, some vendors will face financial woes, depending on their financing, and just not recover. The companies without a unique value proposition will get hit hard in this shakeout, like they always do.

Walker Rowe

Machine learning will be a major 2020 IT trend, and will play an increasing role in monitoring systems, including for cybersecurity. This will reduce legwork and increase accuracy as it keeps people from chasing down events that are not statistically significant. It reduces the signal-noise problem. But there is one giant limitation: the lack of what is called labeled data. You can flag statistically significant events and graph trend-aware data with clustering and other techniques. But in order to identify, for example, data theft, you would need a picture of what that looks like; this kind of deep learning still largely requires that humans look at events and manually key in which combinations of metrics, or features, predict system outages or flag hacking events.

It's like feeding X-rays into neural networks to identify who has an illness and who does not. You need data to train the algorithms. Some companies hire people on the Amazon Mechanical Turk market to manually do that by, for example, looking at photos and identifying people, cats, flowers, etc. But how can we feed in complex metrics like traffic surges and login anomalies and label those accordingly? That would be possible if vendors came together on this effort. For now, it's a big gap.