For readers who would like to see what happened to our 2017 predictions, please take a look. Jens Soeldner and the EMA Team sat down again this month to look into the crystal ball for 2018.

Central Theme for 2018: Transform into a Digital Attacker

All 2018 trends contribute to transforming enterprises into digital attackers that release faster, cheaper and at higher quality than the competition. Digital attackers have the ability to quickly and cheaply try the technological feasibility of  ‘crazy ideas’ and validate these new capabilities with a number of customers.

Digital Attackers Reward Experimentation and Early Validation

This ability to experiment without incurring prohibitive cost is key for digital attackers. While key metrics are key to prevent ‘reckless’ waste of time and resources, a digital attacker rewards initiative and early customer validation.

All 2018 Predictions Focus on this Transformation into a Digital Attacker

Prediction 1 – Machine Learning Will Become Prolific

Today, machine learning often is reserved for specialized programmers and data scientists. These job types are scarce in the marketplace and therefore only accessible to corporations able to pay top dollar for talent. But even these organizations struggle with the fact that their machine learning competences are separated from the rest of their development and data center organizations. Full stack developers and business analysts are much closer to the business on a daily basis and would be able to contribute potentially ‘life changing’ ideas due to this closeness.

Offerings such as AWS SageMaker or AWS DeepLens, but also many much lesser known products aim to abstracting away the complexity of selecting, provisioning, training, testing, and deploying machine learning algorithms and frameworks. Developers receive an API, upload their files and the system takes care of the rest, including the much feared overfitting of algorithms. In 2018, we expect that every full stack developer and business analyst will have access to machine learning capability and will start experimenting with how to leverage machine learning for faster releases, more relevant features, cheaper operations, better performance and availability.

Prediction 2 – Machine Learning and Artificial Intelligence Must Show Quantifiable Value

The days of ‘everything machine learning is cool and worthwhile’ will be over soon and they will be replaced with the need to ‘shift left’ and proof the customer value, cost reduction or speed increase that ML / AI brings. It remains true that every discipline in development, operations and business can benefit tremendously from ML / AI, but to make this happen organizations must show the same discipline in terms of applying metrics, as they are currently showing when pushing hard for DevOps value chain management.  

Successful ML / AI vendors in 2018 will focus as much on showing the limits of their solutions as they will push to show its immediate value. At the same time this means that we will see more ambitious and broader approaches, with less manual input needed. As long as we are open about the obviously needed holes of these solutions, we will jointly find innovative solutions to gradually close these holes while preserving the universal character of these software solutions.

 

Prediction 3 – Serverless Will Become Big for Containers and Beyond

While some vendors in 2017 have attempted to convince you to lift and shift legacy apps to containers due to density, security or portability advantages, in 2018 we will discontinue this misguided approach and focus on why digital transformation is so important. Container solutions in 2018 will be all about providing digital attackers with the speed, scalability and cost savings they need to serve customers better and cheaper.

To serve customers better and cheaper, enterprises cannot rely on expensive staff to manage their Kubernetes stack with all of its integration points to their traditional IT and dev groups. Instead, vendors will look to offerings such as AWS Fargate serverless containers or to the growing number of hybrid frameworks for serverless functions.

 

Prediction 4 – The Docker Company Will Start Focusing on Revenue

Docker only makes money when enterprises buy licenses for the Docker Enterprise Edition Product. In 2017 the company had lost its way by strongly focusing on selling to operators with a ‘lift and shift’- message. After the CEO change the company is aware of its limited runway and focusing again on bringing together developers and operators with a ‘best of breed DevOps’ message.

 

Prediction 5 – RedHat and VMware Are Facing off

RedHat (OpenShift) and VMware (PKS) have the unique opportunity to offer their customers one unified management platform for everything from bare metal to serverless computing. Both companies will throw the ‘kitchen sink’ at this ‘unified management’ story to hand back central control to operators while providing developers with an undiluted Kubernetes platform. If RedHat and VMware are able to execute on this unified strategy, Docker and other dedicated container management platforms will struggle making the argument for a separate container management stack. However, if RedHat and VMware cannot execute, this will jeopardize their market position.

 

Prediction 6 – Filling in Gaps in the DevOps Process: No More Missing Artefacts

Compliance and security concerns are the driving factors behind enterprises no longer putting up with automation gaps and missing artefacts in DevOps. Batch jobs, databases, machine learning models, data from legacy applications, and everything else will have to be integrated with the DevOps pipeline.

 

Prediction 7 – Joint Monitoring and Management for Dev and Ops

Monitoring platforms need input from development in terms of relevant code changes, their user impact, and also their importance to the business. Whenever new capabilities are introduced by development, operators need to clearly understand the operational importance in terms of performance, availability, scalability. In 2018, we will start seeing unified DevOps management and monitoring solutions that enable operators to derive tech requirements directly from the relative business importance of applications and microservices.

Vice versa, in 2018 developers will receive feedback on the operations and end user impact of their code. When a new version is pushed, they receive a scorecard for their new code in terms of operational impact regarding cost, speed, scalability, user experience, and so on. This means that in 2018 developers will be rewarded for releasing code that’s simple to operate and operators will be judged by their ability to not slow down developers while still enforcing compliance and security.

 

Prediction 8 – Finally We Will See Real Private Clouds

In 2018, private cloud is coming back with a vengeance. Virtualization vendors and hyperconverged vendors will lead the charge, offering more and more turn-key EC2 type clouds with SaaS, PaaS, serverless, containers, VMs and bare metal capabilities. This is needed as over 50% of apps currently are not suitable for the public cloud, but enterprises are pushing to still transform them into native cloud applications. This also means that successful vendors will provide a ‘public cloud compliant’ private cloud environment that enables shifting applications between data centers and clouds.

 

Prediction 9 – Service Meshing Will Take Off

The race up the stack will continue so that vendors that make it simpler to mesh up microservices will own critical differentiation potential. Microservice to microservice networking will become critical to enable intelligent microservice to obtain information and capabilities from wherever they are available in a secure and compliant manner.

 

Prediction 10 – The Role of Configuration Management and IT Automation Will Change

In 2017 traditional configuration management has come under pressure through immutable container infrastructure, where containers are replaced instead of reconfigured. What is needed in this world of immutable microservices is a centralized guardian of compliance, security, performance and availability that is independent of whether an app is containerized or still hosted on bare metal. Someone needs to track what can be part of container images, which location they can be downloaded from and what is the impact of deploying a new image or pulling images due to security impact. This means that in 2018 we will finally see a declarative approach to enterprise IT, where requirements are managed separately from scripts, recipes and deployment instructions in general.