Ensure Oracle cloud readiness with scalable, codeless test automation
Watch this webinar learn how you can use codeless test automation to develop a scalable Oracle testing strategy that works before, during, and after the migration.
What will performance engineering look like in 10 years? How will our job change? On a recent Performance Advisory Council (PAC) Online Meet-Up, over a dozen performance testing experts from around the globe weighed in on the future of performance engineering.
The group was universally optimistic about where performance engineering is headed — and about how performance engineers fit into this brave new world. But everyone agreed that the role of performance engineers will change. No matter what the technology, things are always bound to go wrong — and we’ll need people with expertise to fix them.
The most common observations revolved around technological seeds that have already been planted. We’re looking at an amplification, or fruition, of emerging trends, practices, and processes that are already under way today. What’s leading-edge today will be mainstream in ten years.
We asked our experts to gaze into their crystal balls, and here are the top 3 predictions.
To envision where we’ll be in 10 years, think about where we were 10 years ago. Pretty simple architecture, with a couple of layers. Even with all the metrics we were looking at, we could understand everything. When we thought about load testing, all of the modeling was done inside our head. The tools just helped us apply the solution.
But today the human brain can’t process modern architecture with hundreds of nodes, tons of servers, thousands of inter-dependencies. The problem will only get worse and worse: architecture will get only more complex, with more stuff to deal with. If we want to be effective, we’ll need something “smarter.”
Will AI allow us to model the likely performance of an app and its hosting infrastructure to an accurate enough degree that we won’t need to actually run a performance test? Self-maintained or -created scripts, self-analysis, self-defect tracking and reporting?
We have almost all the data to enable us to do performance modeling today – way better than we were ten years ago – because most things are monitored and tracked. It’s just a matter of building it. A few years ago we’d say it was impossible. But now we can easily foresee a central “something” that pulls all the specs and all the telemetry from everything that’s known and then automatically optimizes things behind the scenes without our ever knowing. We already are seeing automated web performance optimization where “why slow?” rules scan your website and tell you what’s wrong with it, tell you your next steps, and then can implement them for you.
Software development will be more like putting together components that are already known and proven. IT will provide these functions — these components — that work very well, then everyone can go off and develop their own stuff. Performance engineers will be the ones who are skilled at integrating and understanding all the different components so they can be optimized.
Performance engineering will move more toward configuration (and tuning) than testing — particularly for a specific infrastructure (AWS, Azure), with app layers on top. There will be AWS performance experts, Azure performance experts, and so on. Scripting will be a thing of the past. How to build a good application with certain components, solve problems on specific platforms or products, set up architecture, etc., will be performance engineers’ job.
We need to up-level our skills in the architectural space and the data science space. As businesses moving to the cloud becomes ubiquitous, performance engineers won’t have their own data to work with. They’ll need to know how to formulate a hypothesis, how to go through evidence, reformulate their hypothesis. The future will be about collecting telemetry to find underlying problems — the right hotspots, meaningful KPIs — and communicate that information in a way that provides actionable intelligence to business and delivery teams.
There will be shift from human beings to AI. Humans won’t be the ones creating tests or analyzing, them. Applications will be able to understand automatically how to test themselves and how to auto-remediate, so the performance engineer will become more an evangelist working with Dev teams, steering them in the right direction from a performance perspective.
It comes down to fitting into a DevOps kind of culture. The testing we did as performance specialists will increasingly be done by delivery teams. Performance engineers will be more enablers than deliverers.
Learn more about the PAC program and its events here.