Harness, at its {Unscripted} 2020 conference today, announced its plans in the fourth quarter to make available as a beta a module that leverages machine learning algorithms to optimize build and test cycles on the Harness Continuous Integration (CI) Enterprise platform.
At the same time, Harness is adding a beta of a Continuous Features module to enable DevOps teams to employ feature flagging to simplify the application development process. A visual dashboard for feature verification and analytics monitors all flags, including active, inactive, live and unrequired flags, as well as customer issue tickets, to simplify the management of new capabilities being added to an application.
Harness is also updating its continuous delivery (CD) platform to include a revamped user interface, GitOps and pipelines-as-code processes, bi-directional synchronization and conflict management, and templates to standardize deployments across teams.
Finally, Harness is adding a Harness Next Generation Continuous Verification tool to provide visibility into the impact changes have on IT environments. This tool can be deployed standalone or used in conjunction with any CD platform and is intended to make it easier for DevOps teams to discover the root cause of issues that typically stem from recent changes made to an otherwise stable IT environment.
Harness CEO Jyoti Bansal said artificial intelligence (AI) in the form of machine learning algorithms are poised to play a major role in taking DevOps automation to the next level. The Harness AI module can reduce test cycle time by up to 75% by correlating and isolating tests to changed code rather than requiring all tests to be executed with every change. The Harness approach means only relevant test cycles are run, he said.
Continuous Integration Enterprise will also be able to identify gaps in test plans by highlighting code changes that are not covered by existing test cycles. Library caching and container image size optimization will also be added in the future to make builds more efficient.
Harness also announced its intent to apply machine learning algorithms more broadly. Algorithms will enable DevOps teams to control, optimize and accelerate the flow of software pipelines rather than requiring engineers to manually verify builds, tests, deployments and releases. Harness will also apply data science techniques to observe the quality, performance, cost and reliability of code as it flows through each pipeline step and, if need be, automatically abort or roll back the delivery of software.
As each build is committed, the opportunity to collect even more metadata about the workflow process only increases, Bansal added.
Obviously, Harness is not the only provider of a CI/CD platform looking to apply machine learning algorithms. Thanks to machine learning algorithms, the rate at which applications are developed and deployed could increase by several orders of magnitude in the months ahead. The issue now is to what degree the rest of the business will absorb that potential rate of change.
