2019 is almost gone—and what a year it was! But now, it’s time to predict the year ahead. What can we expect in the world of infrastructure and application development in 2020? How will DevOps, the cloud, AIOps, serverless, Kubernetes, and programming languages like Java and Rust evolve over this next year and into the future? We asked eight New Relic experts to peer into their crystal balls, and here’s what they had to say:
Tori Wieldt, Senior Solution Marketing Manager
No-Ops will finally die. In 2020, organizations that haven’t already done so will ditch the notion of “No-Ops” and start to recognize the value in their Ops team members. Companies will always need Ops people, but as technology and automation processes mature, Ops professionals will be able to work on higher-quality problems, such as removing friction for their developer counterparts, improving long-term infrastructure scalability, and teaming with software architects to build reliability into new services and features.
Global regions that have been slow to adopt DevOps will finally get on board. Most companies in the U.S. are aware of DevOps. Plenty have pockets of DevOps teams, but far fewer companies are running finely tuned DevOps practices across their entire enterprises. In that regard, we still have a ways to go in terms of mainstream adoption.
Outside of the United States, DevOps momentum is slower but will pick up the pace. In unpublished data taken from a survey of 200 UK IT decision-makers, only 10% of respondents said they have no intention of using DevOps, and less than 1% said they had tried DevOps but stopped because it wasn’t for them. More companies will start to understand the value of elite DevOps performance and will introduce DevOps practices into their organizations.
Guy Fighel, GM, Applied Intelligence and Vice President of Product Engineering
AIOps tools will get easier to implement, learn, and use in 2020. Too many AIOps tools require a big-time investment until you see value from them, and busy DevOps teams still have to spend too much time onboarding these tools, bringing data into them, and training and educating their teams on how to use them. If AIOps is going to reach the masses, this paradigm can’t continue.
Rather than requiring hours to develop and train models for each new integration point, we expect to see AIOps capabilities leverage a flexible baseline of rules that adapt based on user input and production data over time. DevOps teams will have the ability to add their own logic to the system, as well as to accept or give feedback on automatically suggested logic. In 2020, the AIOps tools that become simple to implement, learn, and use will thrive. The ones that don’t will become obsolete.
AIOps will see increased interest from IT leaders. Companies are increasingly deploying artificial intelligence (AI) across departments to help facilitate their digital transformations. According to IDC, global spending on AI systems is expected to climb to $77.6 billion in 2022—more than triple its spending forecast for 2018. While AIOps is a small piece of the AI trend, we expect IT leaders to invest more in this technology to reduce downtime and directly impact the bottom line. IT leaders recognize the long-term benefits of AIOps to support the efforts of DevOps, service desk, and InfoSec teams. As this technology matures, different AIOps tools will cater to these groups, with the biggest distinction being what kinds of data the tool can ingest and how fast they can deliver value.
Andrew Tunall, GM of New Relic Serverless and Emerging Cloud Services
Businesses will learn how to use serverless to optimize for their busy seasons, rather than over-provisioning 365, 24/7. Most businesses have cyclical customer traffic, typically measured by how customers cycle through daily, weekly, or annually. Today, autoscalers are rarely tuned to the right traffic dimensions (or they’re not tuned at all), so teams over-provision to avoid failure for their most extreme traffic use cases. When you look at real-world workloads for most digital businesses, there’s a major opportunity to take things from overweight infrastructure to different, more flexible programming models like those that run in serverless environments.
Functions-as-a-service offerings will evolve and get more capable. To avoid cold start times with functions-as-a-service, one cloud provider recently added a feature allowing you to provision concurrent executions, essentially guaranteeing execution start times within milliseconds. Another provider increased the maximum execution time of a single function to an hour (vs. the industry norm of 15 minutes), opening up the possibility of more complex workloads.
As functions-as-a-service offerings try to cater to more diverse workload sets, I think we will see more of this—capabilities to make functions more resource heavy and longer-running, but still compatible with the event-driven functional execution pattern. So even if you have demanding latency requirements, you will be able to use this technology, taking advantage of the benefits to uneven workload volume, such as customers in industries like retail and gaming.
Lee Atchison, Senior Director of Strategic Architecture
2020 is the year competition to Amazon Web Services (AWS) gets real. Until recently, AWS’s growth has been greater than all of its competitors’ growth combined. That will change this year. In particular, Microsoft Azure will become a very strong cloud option. Azure has a relative stronghold in Europe, but that will extend into the rest of the world, especially in North America. Google will also make inroads in 2020, but it’ll be at a slower rate than Azure. Additionally, expect to see more consolidation in the industry, as smaller cloud providers will merge with, or be absorbed by, these larger cloud players. This phenomenon will be especially true in machine learning, AI, and data management areas, and possibly in DevOps management areas as well.
Finally, companies will start second-guessing their “all-in-on-the-cloud” strategies. There will be much more focus on the balanced use of the cloud, along with private data centers. And, companies' top considerations will include cost and expertise.
Ben Evans, Java Guru and Principal Software Engineer
Expect continued, gradual progress on solving Kubernetes’ complexity problems. As it continues to be a critical project for many organizations in 2020, increased awareness of the complexity of managing Kubernetes at scale will permeate the market. Very few companies will move away from Kubernetes (for a variety of reasons, including the so-called "IKEA effect"); however, companies will look for additional tools and techniques to improve their understanding of their Kubernetes deployments.
Engineers with deep knowledge of Kubernetes will continue to be in very high demand, and companies will increasingly look to implement observability to give their engineers building on Kubernetes the insight they need.
We’ll see the increased growth of OpenJDK. The dynamics we've seen this year in the Java space will continue to play out. The Java market will continue to be overwhelmingly dominated by Long-Term Support (LTS) releases, led by Java 8 (although the Java 11 LTS release continues to gain market share at the expense of the earlier release).
Since Oracle no longer supplies free support (or security updates) for either LTS release, enterprises will continue to migrate to OpenJDK vendors. These distributions provide free (or low-cost) updates for both LTS versions, and adoption of these distros will continue to rise significantly throughout 2020.
Rust will continue its slow, steady rise. 2019 was a year of quiet breakthrough for Rust. Increased support from companies like Microsoft and industry luminaries, such as Bryan Cantrill, has led to a much broader awareness of Rust as an increasingly viable replacement for C++.
The limited deployments we've seen this year have, in my opinion, proven the case for Rust in much wider usage. In 2020, I expect to see some major Rust projects announced—both greenfield and rewrites.
The state of statelessness will finally be observed in 2020. Serverless platforms enforce ephemerality and immutability in a way that demands a stateless architecture. While these are not new, the unimportance of underlying infrastructure reinforces the need for greater application observability.
Legacy enterprises will become serverless innovators alongside cloud native companies. While serverless has quickly gained traction with cloud-native companies, developers in traditional industries like manufacturing and shipping are increasingly skipping containerization in favor of serverless for new projects. For these "leapfroggers," unified observability across legacy deployments and new serverless applications has become a top priority to avoid migration pains.
Joel Worrall, Director, Open Source and Developer Advocacy
2020 will be a breakthrough year in end-to-end observability solutions. For a while now, it’s been possible to monitor most compute processes and systems that make up a value chain, from the point of purchase through the backend. However, there’s never been a way to connect all that data together in one experience. 2020 will be the year someone in our industry realizes that promise. And it will rely on open source software.
Generous partner ecosystems and highly motivated teams will win the observability space. In an age of vendor lock-in, tool consolidation, and solution convergence, the observability platforms that create value for everyone will counter predatory platforms and business relationships.
Additionally, the true innovations in observability will be the result of small, highly motivated teams. Platforms/solutions that support small, focused teams (or even individual actors) will be the genesis of the true innovations in observability solutions.
Of course, as we all know, the future is uncertain. The biggest tech story in 2020 may be something nobody ever saw coming. What’s your tech prediction for the new year? Share with us @NewRelic using hashtag #techpredictions.
Get a free trial of New Relic One here.