I'm the author and mantainer of several Go backend systems. One of them, implements a synchronization layer between the corporate and the legacy relational database engine (Oracle) and the NoSQL (Redis) persistence layer for the new availability calculation engine on the cloud. This service, synchronizes about 40 million of records dayly. With a synchronization time below 1 second in the 99% of cases, and with a 0.0001% of error margin. During the peak season, this traffic may rise to 100 millions of synchronized records by day. To do this, the desing was made with reliability and performance as key features, taking advantage of the awsome concurrentcy capacities of the Go development platform as buffered channels and go routines.
On another hand, I have worked with other systems also, as a and extends auxiliar library with util components as a generic sharded pool, a timeout wrapper, or a very basic components container.
In addition to Go platform, I have containerized all my components by using Docker, in continous delivery scenario using Jenkins.
Another very important work in Go, was a the implementation of the Go client for de BPulse monitoring system (bpulse.io), made it in collaboration with the BPulse team. You can view this project at my GitHub profile (github.com/theskyinflames)
I have designed and implemented all of components that conforms the synchronizaton layer following the "maximum decoupling and maximumly cohesive" rule.
To decouple the components I have used Kafka as sequenced messaging broker.
As a VCS I use Git.