SLEEP SOUNDLY WHEN YOUR CODE IS RUNNING ON MILLIONS OF DEVICES

 

 

After two years, Heapcon, the biggest IT conference in the Serbian market back in live. Heapcon, traditionally organized by Heapspace gathered the biggest names from the tech industry, a lot of famous international speakers, and more than 20 IT companies supported this event. Visitors could enjoy various topics and talks, but meet the companies as well, try some little challenges, won some prizes, and have a lot of fun and networking. 

 

This year, United Cloud supported Heapcon as a Gold sponsor, but more important was our story about soundly sleeping. Our awesome guys and senior engineers - Boris Petrovic and Miroslav Mijajlovic shared how we sleep soundly when our code is running 24/7 through millions of devices. This was a unique talk in a completely full U track room, where we shared what we've learned through these couple of years and the sleeping was not soundly always. Going through horror stories and unslept nights they explained what engineering practices and corrective measures we implemented on the way to better sleeping. Better metrics from performance testing and reproducing real traffic to test environment were just the beginning. Boris talked about the importance of monitoring and alerting and why the self-service is only right way to ensure alerts, metrics, and quality of code.

 

“The concept of the self-service model is based on the idea that you can solve any problems independently,  without engaging other teams or interrupting your own workflow. By doing so, you can increase speed and reduce the number of context switches, which ultimately leads to improved productivity. This model most people use frequently in activities like shopping and booking flights online.” 

We made a lot of improvements, but we still had some troubles in deploy pipeline. Miroslav has shed light on long-living branches and merge hell. This is something where a lot of engineers can find themselves. He discovered what benefits we got from trunk-based development and why feature toggle is an important thing when you are working on such a large scale.

The story about soundly sleeping has spread around the Madlenianum Theatre. A lot of visitors, fascinated by this talk were coming to the UC booth to find out more or just to share their own stories. Furthermore, they could start “hunting for treasure” and the best hunters were prized by the shirts for better sleeping.

"I really find this talk useful, I have a problem with the nightly updates...", and "Maybe the most useful talk at this conference, thank you guys!" were just some of many comments and impressions.

We are so proud we can share a unique story of what does look like when your code is running 24/7 on millions of devices. We are proud we can say we have unique skillsets and engineering practices, but we can easily share all of our fuckups and failures at the same time. 

Hope we'll see again next year and until then - sleep soundly, and don't let bedbugs bite! :)

 

Sign In