Previous proposalUsing Powershell & Teamcity for Windows based Deployments on AWS
Next proposalDevOps and QA
Good to Go
Submitted by Arpit Mohan (@arpitmohan) on Thursday, 28 January 2016
Section: Full talk Technical level: Beginner Status: Submitted
Coming from a traditional LAMP stack, this talk focuses on how & why we decided to re-write our core platform in Golang. I’ll also stress upon the learnings and heartbreaks through this journey.
Traditionally, any startup looks for the quickest way to reach the market. This generally involves writing the first cut of the application in a language that is familiar to the founders. A few years back, this meant that most startups began their journey on a monolithic LAMP stack. As the startup begins to grow, the monolithic codebase no longer works. The team generally decides to move to a distributed services stack. Then, at scale you figure out that there are bottlenecks in your codebase. Also, there may be gaps in your monitoring system, deployment process and ability to handle operational issues.
This is exactly what we, at Exotel, also went through as we grew from 10 calls/day to 4 million/day. This is a story about how & why we transitioned from a LAMP stack to a Golang powered microservices architecture. The talk will revolve around the tradeoffs we made, how & where we failed, what we learnt (specifically in monitoring & downtime response) and what we are doing to make life better for ourselves.
As a veteran of 2 startups I have a plethora of failure stories behind me. Currently, I’m a senior engineer at Exotel tasked with the responsibility to build a scalable, future-proof platform. I enjoy tinkering around with new software & technologies.
- INK Talks https://www.youtube.com/watch?v=c7EDpnagEZI