Reliability: Never Roll Out to All of Prod at Once

Alexey Gorbov works on databases as a service: Cassandra, Postgres, caches, and platforms must be convenient for internal users. His interest in systems grew from books, mathematics, independent study, and Debian; alongside it he administered a school, a pharmacy chain, and a research organization.

At a telecom provider, Alexey built resilient mail, DNS, and hosting before Docker. At Odnoklassniki, generalist administrators owned the whole service. A three-day outage after a faulty rollout exposed the cost of speed without safeguards and led to runbooks, gradual releases, and reliability work.

His expertise grew from operating Cassandra, Voldemort, and distributed caches. Adoption combined rules, proven value, and agreements with teams. Moving a data center in two months without downtime required hauling servers, planning networks and order, recovering failed hardware, and coordinating people.

At T-Bank, Alexey improved hiring and candidate-data protection, inventoried clusters, reviewed incidents, and helped with data models. Cassandra became a push-button service and a proving ground for other databases. His advice: understand a technology, test it immediately, and read widely, including fiction for language and intuition.