They had two backups. Both died in the same outage.
Team felt safe: primary server plus a backup server. Both lived in the same cloud region, same availability zone. A zone-wide network partition took both offline for 14 hours during peak. "Two copies" meant nothing because both shared one failure domain.
No data lost — but 14 hours of downtime and a churned enterprise client that needed 99.9%.
Two copies in one place is one copy with a typo. The 3-2-1 rule says different media AND a different location for a reason. Fight me.
Backup or Die
@BackupOrDie
They had two backups. Both died in the same outage.
Этот пост опубликован в Telegram-канале Backup or Die. Подписаться можно по ссылке: @BackupOrDie.