## Description Formulate hypothesis, conduct investigation and provide evidence that Redis sharding would help to improve reliability and fault-tolerance of large GitLab installations. We want to proceed forward in three directions: 1. Make current Redis architecture more resilient - same version of Redis in development/staging/production environment -> https://gitlab.com/groups/gitlab-org/-/epics/2661 - testing Redis failover on staging using chaos endpoint -> https://gitlab.com/gitlab-com/gl-infra/scalability/issues/133 2. Improve observability / predictability - Record Redis Persistent requests timing -> https://gitlab.com/gitlab-com/gl-infra/scalability/issues/157 3. Unlock progress on sharding in case we will need it - Redis CROSSSLOT validator -> https://gitlab.com/gitlab-org/gitlab/issues/206903 * Google Doc with notes from calls about Redis improvements :arrow_right: https://docs.google.com/document/d/1QXZseUhZweB_Fl4HsOoIj56bdF1w5T74k0j5ELy6o-8 * Google Doc describing our strategies for Redis sharding on gitlab.com :arrow_right: https://docs.google.com/document/d/1vKcT_PcX6qDHeWTA-e4we9eambKQPq6SGHnRLYN_1GM