Will automation replace developers
|Richard Harris in DevOps Monday, April 24, 2023|
We recently caught up with Richard Whitehead from Moogsoft and asked him all about automation including if it will take over developers' jobs, the differences between full and partial automation, some barriers that organizations face when trying to implement autonomous processes, and much more.
Richard Whitehead discusses automation, the steps your organization can take to implement autonomous processes and the main barriers companies face, what role full and partial automation will play in the future, some critical skills IT managers need to have when working with automation, how you can improve availability and support with AIOps, and more below.
ADM: Do you envision a future where everything is fully automated? Why or why not?
Whitehead: I don't. I like to say that full automation is something everyone wants, but nobody does. This is because organizations using full automation can go from a slight mistake to a catastrophe without a human ever knowing what's happening, and that's scary. So, given that IT systems are complex and fragile, many teams want to intervene or be the final stop. In an SRE environment, engineers may restrict automated code pushes to protect error budgets. Partial automation is a more achievable goal than full automation and should be a top priority for tech leaders. In the AI space, we refer to this as "human-in-the-middle."
ADM: We often hear the fear of automation taking over jobs. Is that a valid concern? Why or why not?
Whitehead: Automation taking human jobs is not a concern we should worry about right now. Businesses should recognize that instead of replacing IT jobs, automation supports the IT team. Automation takes repetitive tasks off of SRE teams' and DevOps practitioners' plates, freeing time for them to focus on more valuable tasks. For instance, automated solutions, like artificial intelligence for IT Operations (AIOps), can scan mountains of data and detect potentially service-impacting incidents before they impact end users. Without AI, that would be like finding a needle in a haystack, impossible, even for the best-resourced teams. But even this technology doesn't replace human teams. Rather, the platform notifies teams when there's a problem in the IT ecosystem that needs to be fixed with human ingenuity and creativity.
A recent Dell study indicates that organizations are advancing in digital transformation yet lagging in automation adoption. As IT environments become increasingly complex, automation will become increasingly necessary, requiring hesitant companies to adopt automation to keep up with digital transformation.
ADM: What holds organizations back from implementing autonomous processes?
Whitehead: Organizations are hesitant to implement automation for a simple reason: people are nervous. Change is difficult, and people crave human oversight. Even in a fully-automated world, people still like notifications and a button to push. But beyond nerves, additional concerns preventing automation include:
- Trust: Teams need proof automation works before trusting the system to do the right thing.
- Budget restrictions: Implementing automation can be costly, and the process has to be frequent to justify the expense. I believe in the “rule of three”: if it's going to occur more than three times, automate it.
- Lack of in-house expertise: Writing automation code can take time and effort to set up. If businesses don't have an IT team skilled in automation, they could struggle to implement automated processes.
- Security: It's essential to ensure that your automation cannot be exploited by bad actors. For example, expensive, time-intensive automation can be a boon to someone wanting to replicate the effect of a DDoS attack. Think of it like Hans Gruber in Die Hard, exploiting a known FBI response to achieve his real objective.
ADM: What is partial automation, and how do you know if that's the best strategy for your organization?
Whitehead: Partial automation is setting up automation with a "human in the middle" approach. As I mentioned, even DevOps practitioners still prefer to push a button. With partial automation, human operators collaborate with the technology on outcomes, customizing algorithms and training the tech to repeat desired actions. Building a partial automation strategy allows organizations to dip their toes into automation and enables teams to increase their confidence in the automation tool.
ADM: What are the ideal skills IT managers should have when working together with automation?
Whitehead: Organizations need team members that can build code and understand the back end of automation processes. But when it comes to management, it's all about mindset.
To implement successful automation, leaders need to trust their tools and maintain a growth mindset. This is especially important for leaders of small SRE and DevOps teams that can only scale with help from automated tools. Scalability is likely why 36% of organizations with high automation maturity expect revenue growth of 15% or more, according to the Dell report. Comparably, only 10% of organizations with low automation maturity expect that same growth.
ADM: What are some small steps organizations can take to implement automation into their workflow?
Whitehead: Most importantly, focus on getting the IT operations team on board. If reluctance to change creates automation hesitancy, start building trust between the IT team and the technology using partial automation. Ask human teams to train the technology as they would a pet, through positive and negative feedback, steering the tool toward ideal outcomes. Use semi-supervised automation that allows the teams to customize ML algorithms. And, finally, consider adopting tools that allow users to peek behind the curtain, see how the technology works.
ADM: Automation often leads to improved availability. How does AIOps support automation and availability?
Whitehead: Two critical features of AIOps support automation and availability: ML and natural language processing (NLP). AIOps solutions identify data anomalies that indicate a potentially service-impacting incident and alert SRE and DevOps teams that there’s a problem to address. But AIOps doesn't stop there. The ML feature learns destructive patterns and prevents them from happening again, lessening incidents and outages and their associated unplanned work.
With NLP capabilities, AIOps understand the nuances and flexibility of language. Previously, only humans could distinguish that 10 North First Street was the same as 10 N. 1st St, creating an incredibly complicated infrastructure of rules. But with NLP, code can move beyond a raw match to look at words, understand English, and make the correct correlation.
These features mean that AIOps technology is flexible and robust in the face of change. Automation tools with this kind of staying power are much easier to justify financially than old, brittle automation tools.
About Richard Whitehead
Become a subscriber of App Developer Magazine for just $5.99 a month and take advantage of all these perks.
MEMBERS GET ACCESS TO
- - Exclusive content from leaders in the industry
- - Q&A articles from industry leaders
- - Tips and tricks from the most successful developers weekly
- - Monthly issues, including all 90+ back-issues since 2012
- - Event discounts and early-bird signups
- - Gain insight from top achievers in the app store
- - Learn what tools to use, what SDK's to use, and more