Postgres Glossary
Below you can find beginner-friendly explanations for common terms used within the world of PostgreSQL. We've also included links for additional reading about each topic. Feel free to if you don't see it here.
Our open-source application that collects metrics, queries and other useful data from your Postgres server. Learn more.
Wasted space that accumulates in tables over time due to dead rows not being VACUUMed often enough. Learn more.
Amount of space on disk that is read/written from OS page cache. Typically in chunks of 8kb. Learn more.
The shared buffer cache that Postgres uses to cache table and index data to speed up queries. Learn more.
When a query is served from the shared buffer cache and a slower trip to disk is avoided. Learn more.
Number of times a query is executed. Learn more.
Command to save the current transaction to the database. Learn more.
Modified blocks in Postgres shared buffer cache updated by a query. Learn more.
Estimated row counts are updated by VACUUM processes and due to sparse vs dense pages, the estimate can vary significantly. Not to be used as an exact number. Learn more.
Plan that the query planner uses to execute a query, showing which indexes and tables were used. Learn more.
Replicated data flushed to disk on a replica. Learn more.
Platform-as-a-Service (PaaS) company that offers hosted PostgreSQL servers. Learn more.
Easily installed tools and services that can be added to a Heroku account. Learn more.
Data structure that speeds up query reads at the cost of additional overhead for writes and updates to a table. Learn more.
Level of work that a server is performing. Learn more.
Postgres connection pooler that supports a higher number of database connections than standard Postgres. Learn more.
Personal Identifiable Information. The agent automatically filters out or redacts any data from queries or EXPLAIN plans that might be PII. Learn more.
Powerful open-source relational database that emphasizes extensibility and SQL compliance. Learn more.
Queries canceled due to conflicts with standby recovery in a database. Conflicts only occur on standby replicas. Learn more.
When two or more queries are blocked on each other due to desired locks being held by the other query. Learn more.
Fully applied replicated data on a replica. Learn more.
Read-only Postgres server that is replicating data from a primary server. Useful for scaling read-heavy databases or for supporting primary failover situations. Learn more.
A row or tuple is an entry in a database table. A database row is similar to a row in a spreadsheet, where the columns are the schema for the table. Learn more.
Queries that do not use an index and scan through an entire table, often slowly. Learn more.
The structure of your table and indexes. Learn more.
Key/value pairs within a query comment. Useful for troubleshooting where a query came from. Learn more.
A collection of related data that is stored in a database. Learn more.
Short-term storage space on disk that is used by Postgres for temporary tables or temporary work. Learn more.
The Oversized-Attribute Storage Technique Postgres uses for storing large fields outside of a table row. TOAST is only used if a row contains large data that cannot be stored within the row's page size. Learn more.
A set of multiple operations or queries that are performed together atomically. Either all complete successfully or all are rolled back on any failure. Learn more.
Write-ahead log used by Postgres for crash and transaction recovery. Learn more.