YugabyteDB features two services: the YB-TServer for database-level operations and the YB-Master for cluster-level controls and metadata storage. Its unique structure divides tables and indexes into tablets, leading to RAFT groups with a distinct leader-follower dynamic based on the replication factor.
Tips and Tricks
Learn how YugabyteDB handles node loss and recovery to ensure consistent, continuous availability. With features like a 3-second leader re-election process and a robust fallback mechanism, your cluster can tolerate disruptions and resume optimal operation efficiently.
Discover why YugabyteDB doesn’t support PostgreSQL extensions pg_repack and pg_buffercache. While YugabyteDB leverages the Postgres query layer, its unique approach to the storage layer makes certain extensions unnecessary.
Discover how YugabyteDB gives you the flexibility to set different RF values and server counts for your read replica cluster. Tailor your replica nodes for optimized low-latency reads across multiple regions while maintaining a separate replication factor for high availability.
Learn how YugabyteDB ensures stability when a downed node rejoins the cluster. Our latest blog covers the essential 15-minute RAFT timeout and the built-in fault tolerance mechanisms that keep your data secure and accessible.
YugabyteDB offers ways to geo-distribute reads. While read replicas are one method, they require more infrastructure. An alternative is to enable follower reads for low-latency access from the primary cluster.
In YugabyteDB, you can manage the number of shards (tablets) using flags like ysql_num_tablets for total control or ysql_num_shards_per_tserver for control per tserver. Dynamic auto-splitting is also supported for automatic adjustments as data grows.
YugabyteDB uses LSM Tree / SST Files to minimize the risk of data corruption, in contrast to PostgreSQL’s method. This approach, coupled with independent compactions and checksum verification, enhances data integrity.
This edition of the Distributed SQL Tips and Tricks blog looks at how to use READ COMMITTED isolation in YugabyteDB to avoid restart errors in YSQL, ensuring a smoother database experience.