For those not familiar, Jobvite is an industry leader in recruiting software that helps emerging, mid-market, and enterprise companies hire top talent easily, efficiently, and effectively. Based in San Mateo, California, Jobvite provides a SaaS recruiting platform for thousands of customers and over 64M+ candidates.
A few weeks back, I got to spend time with John Stuart, who leads the DevOps team at Jobvite. The first thing I quickly learned about John was his unrivaled passion and focus for the customer. Pretty much everything we talked about always linked back to the customer. I’ve met a lot of DevOps leaders over the years, but not many “get” DevOps like John does.
Going Beyond Continuous Integration
It’s fair to say that Jobvite has a mature Continuous Integration platform in Jenkins, which allows engineering to take code to artifact. Over the past few years, this Jenkins platform has been extended, primarily with custom scripts to cover various parts of their Continuous Delivery process. A key challenge John and his release team faced was the growing complexity of maintaining custom scripts as their core application, services, and technologies evolved.
In addition, all performance tests and quality checks were manually done across their development lifecycle. The net result of all this was that CD at Jobvite was a semi-manual process that was complex to maintain, as well as to debug when deployments failed. This meant a six-week release cycle for Jobvite, something which didn’t sit well with John: “I want our developers to start deploying daily and hourly.”
The Continuous Delivery Initiative
An internal CD initiative sponsored by the CTO at Jobvite allowed John to truly partner with his engineering peer so he could provide the automation and tooling required for engineers to deploy their own code and artifacts. When I asked John how his release team felt about this change, he replied: “At first they were cautious, because it felt like I was taking responsibility away from them. However, when I explained that this change would mean developers would fix their own deployments, it absolutely resonated because it would mean fewer fire drills and troubleshooting for everyone on the team.”
The only ask for this CD initiative was “Ensure performance & quality for all deployments, and do so without impacting customers.” Again, the relentless focus that Jobvite has on customers and their experience is impressive to observe. Everywhere you walk in the Jobvite office, you’re constantly reminded of what’s important with posters, visuals, and quotes.
John and his team were introduced to Harness during our beta program. “Harness was easy to install and setup, we were literally up and running within a few hours,” said John. By the end of day one, his team has built and deployed their first CD pipeline. Shortly after, they had built and deployed a pipeline for one of their production microservices, which were built using Java, Docker, Kubernetes and AWS. In addition, Harness was able to integrate with Jobvite’s ecosystem of DevOps tools like Jenkins, AppDynamics, and Elastic.
A clear benefit for John and his team was the ability to perform canary deployments. “Our enterprise customers value stability over features,” said Jon, “With Harness, we can now send the canary down the coal mine, test a new feature or bug fix, verify it, and then absolutely sure it’s ready to deploy to our enterprise customers.” This ultimately translates to a higher quality of service and experience for Jobvite customers.
As well as being able to verify the performance and quality of deployments, Jobvite was also able to reduce their deployment time by 90%. Prior to Harness, a typical production deployment for the team would take 27 minutes; now that same deployment takes just 2 minutes. Jobvite didn’t need to rewrite their existing CD process; they simply reused and automated how their existing scripts, repositories, and Jenkins CI platform performed together.
Moving forward, engineering will be empowered to deploy their own code directly into production using Harness with John’s team providing the required tooling and support. This means developers can build, deploy, and fix their own code with the release team focusing on automation and tooling.
Build vs. Buy – The Harness ROI
Like many customers with a CI/CD initiative, the build vs. buy conversation happened. For John, the decision was clear: “It would have taken us $500,000 to build, let alone support and maintain Harness each year.” In addition, keeping the status quo was also not an option. “Keeping our existing manual process would require 2 full-time release engineers – I would rather reallocate those resource to more business-critical needs. John concluded, “The ROI for Harness is easy to calculate.”
Watch the 2-minute video of John describing his Harness experience: