Page 40 - FCW, September/October 2021
P. 40

Meet the Next Generation of DevOps Practice: Intelligent Software Delivery
Some federal agencies still struggle with iterative software delivery that can power digitization, however, intelligent software delivery can help with that struggle and speed the journey to digital services.
Agencies have been behind the curve in adopting new technologies and at becoming efficient at DevOps and DevSecOps practices and toolchains. Some are holding onto outdated toolchains—much of DevOps inherently consists of some legacy technology—and that can make it more difficult for agencies to turn out innovative citizen-services, or internal applications in a timely manner.
The intelligence agencies have been further along than most in the government at adopting DevOps, because many of their technical people come out of institutions such as Los Alamos National Laboratory and Oak Ridge National Laboratory as well as Silicon Valley companies, said Billy Miller, federal sales engineer at Harness.
“They’re already on the cutting edge of what is going on” with the latest development tools and techniques,” Miller said.
Other agencies haven’t become as proficient in creating simple and secure way for engineering and DevOps teams
to release applications into production, according to some experts.
“They are trying to figure out a
way to get there but I think they’re
still falling short,” said Paul Almeida, vice president of federal sales at Harness, which provides a platform
for intelligent software delivery. The company’s platform uses machine learning (ML) technology to detect
the quality of deployments and automatically roll back failed ones, saving time and reducing the need for custom scripting and manual oversight.
Many federal agencies DevOps efforts, according to Almeida, try to combine the concepts of continuous integration (CI) and continuous delivery (CD) and treat them as a single tool, rather than taking each to a higher level of performance.
New Vision
To improve software delivery processes, federal agencies need to embrace a new vision for software delivery using the latest technological capabilities.
A platform, such as the one offered by Harness, enables self-service deployments. Engineers can also leverage what the company calls Smart Automation to build pipelines in minutes. The platform applies AI and ML to an organization’s existing
monitoring and log data to verify the success of deployments.
Among the key considerations with software delivery are compliance and governance, so a platform should build enterprise-grade security into an organization’s pipelines, ensuring compliance and governance through all stages of development.
Important security features include single sign-on (SSO) and role-
based access control that enables managers to govern teams across the organization; follow audit trails with a catalog of all events; use native secrets management; and secure integration with other platforms.
AI and ML can also be applied to CI. For example, testing is the most time-consuming component of CI pipelines, and AI/ML can decide which tests to run and in which order. Harness calls this “test intelligence,” and said it can lead to 50% reduction in overall CI pipeline execution times with no reduction in quality.
The Harness platform uses containers to standardize and
drop pre-configured steps into an organization’s pipeline. Each pipeline step is executed inside an isolated container, and every build runs in the container so there’s no concern about conflicting builds on shared servers across teams.

   38   39   40   41   42