What pulls me in is the seam between the code and the live systems it runs on. The interesting problem is never the feature. It is what happens when real traffic hits it: the query that goes slow at the worst possible moment, the broker that turns into a single point of failure during a spike, the data that has to stay queryable long after it goes cold.
Lately that curiosity has pointed at real-time analytics: getting large volumes of live data to land fast and stay fast enough to query while it is still arriving.
- Email: connect@xn--uech8c.com
- Twitter: @shethshails
- GitHub: @DynamiteC