Amvera, a PaaS provider, experienced a total failure of its primary Russian region on October 11 due to infrastructure issues at Yandex Cloud. The company had previously prepared for such a scenario by creating independent backups in a separate data center and establishing replication for its PostgreSQL databases.
Following the incident, the team restored the user dashboard by 12:00 on October 11. By 20:30, users gained access to application and database backups. A new region, Moscow_1, was launched by 21:30, a process that originally had a projected timeline of one week.
By the night of October 12, the company had restored 99% of active applications. Additionally, 99.7% of PostgreSQL clusters on high-speed disks and 95% on slower disks were fully recovered. The company began accepting SLA compensation claims from users on October 15.
The recovery process involved manual intervention and the development of custom tools to provide access to data. To manage load, applications were initially restored in a stopped state, requiring users to trigger a rebuild. The company is now working on further infrastructure stabilization and automation improvements.
