Experts' cloud predictions for 2020

Containers

Activity surrounding application containerization has slowed down. The fierce competition between Docker and Kubernetes that made headlines in 2016 and 2017 is basically over, with Kubernetes having won. Yet, a new round of efforts will build next-generation container technologies that can compete with, or add new functionality to, Kubernetes. Kata Containers, an open source container runtime that launched in late 2017, is one example of this prediction already playing out.

Mirantis' acquisition of Docker Enterprise in 2019 -- and the related announcement that Mirantis will eventually cease supporting the Swarm orchestrator -- will drive some stakeholders and other companies who are working to ensure that Kubernetes doesn't become the only viable orchestrator. There are open source components of Docker that companies or developers can reinvigorate and perhaps turn them into compelling alternatives to Kubernetes.

All of this activity could translate into new cloud-based container services from vendors like AWS and Microsoft Azure. Their container offerings haven't received much love or seen major updates in the past couple of years, but that will change if new types of open source container projects continue to emerge or if open source Docker tooling sees renewed investment.

-- Chris Tozzi, IT consultant

Containerization will continue to grow rapidly and reshape the IT landscape but may finally provoke countervailing attention from CIO level and above regarding its direction and management. In other words, the tail has been wagging the dog and despite all the value inherent in containerization, it is not an absolute panacea. It may thus begin to be subjected to greater scrutiny, for example on security grounds.

-- Alan Earls, IT journalist

Machine learning

Not many cloud predictions matter to the general populace, but this one about the power of AI affects everyone. In 2020, explainable AI will rise in prominence for cloud-based AI services -- particularly as enterprises face pushback around the ethical issues of AI. Explainable AI is a technology that provides justification for the decision that it reaches. Both Google and Microsoft have launched explainable AI initiatives, currently in early stages. Amazon is likely to introduce some explainable AI capabilities as part of its AI tools.

Through the power of deep learning, data scientists can build models to predict things and make decisions. But this trend can result in black-box algorithms that are difficult for humans to make sense of. The biggest challenge enterprises face is the need to track bias in AI models and identify cases where models lose accuracy.

Explainable AI techniques promise to identify these problems so data scientists, managers and the users of AI-enabled software applications can contextualize the results of AI models. Explainable AI will also grow in importance as enterprises face regulations like GDPR that require them to disclose how they arrive at various decisions that affect people.

Since 2018, Microsoft open sourced its InterpretML library for explainable AI, acquired explainable AI startup Bonsai and launched model interpretability in Azure Machine Learning service. The key challenges these services from Microsoft, Google and other industry players face lie in identifying the best user experience, so they can present explanations in a way that adds value for diverse people.

-- George Lawton, IT journalist

Security

As cloud security concerns continue to mount in 2020, the major cloud service providers should build out their cloud security practices and offerings. In response to customer demand, AWS, Azure and Google Cloud Platform each will increase their security offerings and services.

The new service offerings may cannibalize some small cloud security startups, which will be acquisition targets for a CSP.

As CSPs -- AWS in particular -- seek new areas of growth, security-focused professional services are a natural move. Such growth could take place organically, with hiring sprees, or through acquisition of small security firms. We might see some awkward moments between CSPs and their alliance partners.

While the cloud shared responsibility model will remain unchanged in 2020, some security-related changes are on the horizon. Breaches at major companies remain fresh in decision-makers' minds. Some new services and offerings will slice and dice the model where a CSP takes on a larger share in the security responsibility -- for a price.

-- Will Kelly, technology writer

IaaS market

Famously, Casey Stengel, the New York Yankee's manager from 1948 to 1960, warned people to "never make predictions, especially about the future." Nevertheless, the pull of cloud predictions is irresistible this time of year, so here are a few that might come true in 2020.

There will be at least one major enterprise security breach involving lost data and/or disrupted applications on a major cloud service provider. While cloud critics will use the breach to condemn the entire cloud model -- unlike the Capital One incident -- this one won't involve an ex-insider with detailed knowledge of cloud operations. It will expose a fundamental misunderstanding of the shared security model in which cloud operators secure the infrastructure and core services, while cloud users must secure their resources, such as instances, networks, databases and storage. The incident will lead at least one major cloud operator to change its policies, to automatically run security scanning tools. The cloud providers already offer these native tools, such as Amazon Inspector and GuardDuty or Azure Security Center.

In its continuing struggle to keep up with AWS and Azure, Google will extend the Anthos application platform that provides consistency between on-premises and public cloud deployments to a broader range of hardware platforms that run in more places than just enterprise data centers. These will include hyperconverged infrastructure systems targeting branch offices, retail and other remote sites, and embedded systems, i.e. IoT devices. It will also extend Anthos beyond Kubernetes compute and related container services to include other cloud services such as a database, serverless, object and file storage and a subset of data analysis features.

-- Kurt Marko, consultant