In 2011, Sylvain Kalache was a software engineer at SlideShare, which LinkedIn acquired in 2012. Kalache was a senior systems reliability engineer (SRE) at LinkedIn, doing project management and planning, documenting and sharing knowledge tools, and adapting himself and his work to the LinkedIn organization and its technology.
Today, Kalache is helping build DevOps careers as co-founder of Holberton School of Software Engineering in San Francisco, where students learn to code through “project-based and peer learning,” according to information from the Holberton School website.
This is Kalache’s DevOps career story, how he worked in and then transitioned from a career in DevOps to a personal and professional investment in helping students to build their DevOps talents toward a lifetime of coding and contribution in software engineering and, hopefully, successful careers.
Software Engineering at SlideShare
“SlideShare was a very small company when I joined. The San Francisco office had only 10 people; I was only the second engineer at that location,” says Kalache. But the DevOps mindset was strong there, he says, and the company had bet everything on DevOps methodologies. In fact, it was the commitment to DevOps by SlideShare and its CTO at the time, Jon Boutelle, from which Kalache learned of the development approach and took to it so quickly.
“Boutelle was looking for full-stack or aspiring full-stack engineers and I was a junior programmer and eager to learn. In a small company like that where you are encouraged to work on every aspect of the product life cycle, it became very clear that I was in charge of taking care of my mistakes,” says Kalache, adding SlideShare expected him to take ownership of his code from design to production.
“Because I was eventually in charge of a lot that was going on with the infrastructure at SlideShare, I had to build and maintain a lot of systems that were serving billions of slide views per month at what was one of the top 100 most visited websites in the world,” says Kalache.
People from the Dev and Ops teams there did not only their own work, but also the work of their companion team. In essence, Dev was both Dev and Ops and, likewise, Ops was both Ops and Dev.
“The SlideShare people were writing their own configuration management code, setting up monitoring of new services and features as they built them, deploying their own code, and troubleshooting code in the production environment,” says Kalache. Ops people helped developers design sound code for acceleration while also giving their code a second look to ensure fitness for operations.
Project Skynet
“The most exciting project I built for SlideShare was Skynet, an automated tool to monitor, scale and auto-heal a system in the cloud,” says Kalache. Skynet was developed to meet several challenges in the day-to-day work of simply being SlideShare and doing what SlideShare does. SlideShare acted on, renamed and translated countless files from one type to another to share them on the site. As with other slide technologies, the tool was a fit for teaching and presentations.
There was no way to know exactly how much users would upload, access or view any slide content at a given time. “There were peaks and valleys in upload traffic with mornings seeing high traffic and low volumes in the evening and on weekends. The SlideShare infrastructure had to scale quickly to encompass and convert common document files at speed. This required adding many hundreds of machines to the infrastructure,” says Kalache, which were virtual machines that spun up in minutes in the AWS EC2.
One purpose of Skynet was to enable precise analysis of log files for transparency into everything from individual file conversions to changes in the entire machine cluster as the environment scaled up or down. The other purpose of Skynet was to use the analyses to auto-heal; in other words, to restore the cluster back to its initial size once the peaks were over.
With many thousands more employees at LinkedIn than at SlideShare, Kalache left many of his duties at SlideShare to others while he dove much more acutely into specific aspects of his work.
Co-Founding a School That Builds Careers
In speaking with his colleagues and representatives from various enterprises, Kalache continually confirmed that organizations using DevOps are in need of lots of engineers but they can’t find candidates who are knowledgeable and experienced. “I was actually recruiting for SlideShare and for LinkedIn. What saddened me was to see a graduate who had just spent half a decade and a lot of money on a master’s degree only to be far from employable,” he says.
Kalache found it hard to find people who knew basic software engineering, DevOps, and how to be an SRE for LinkedIn. “We instead had to hire early graduates who had coding skills and an interest in operations and train them on being an SRE,” he says.
Kalache decided to combine an awareness of alternative teaching methods with his passion for the coding and ops communities to find a solution to how to better prepare software engineers for careers. “After mulling it over for two years, I decided to quit my job at LinkedIn and create the Holberton School,” says Kalache.
In Part Two, we dive headlong into the teaching and learning experiences at the Holberton School of Software Engineering to help students learn and understand the DevOps methodologies and increase their chances for success in their respective careers.
