Data consistency

In my first engineering role in a software call center company, we wanted to add fault tolerance to the existing solution. After fruitless attempts to invent a consensus algorithm, we switched to ClusterLabs’ Pacemaker. Several years later, at Yandex, I skipped the invention step and was the first in the company to bring in Zookeeper and use it to build a cluster configuration management system. Later, I used etcd (the backbone of Kubernetes) and contributed to the open-source, improving etcd’s update performance. This…