Definitions of Vector-specific terms


A batched payload within a sink. It is a batch of events encoded into a payload that the downstream service understands.


A test designed to measure performance and resource usage. You can learn more about Vector’s benchmarks in the Vector repo’s main README.


The static binary that Vector compiles to.


An ordered queue of events that’s coupled with a sink.


An umbrella term encompassing Vector sources, transforms, and sinks. Vector components can be assembled together into a Vector pipeline for processing observability data in a flexible and configurable way.


The settings and options used to control Vector’s behavior. You can learn more about Vector’s configuration on the Configuration page.


The ability to retain data across exceptional events. In the context of Vector, this typically refers to the ability to retain data across restarts.

Enrichment tables

File-based information that you can use to enrich Vector events in remap transforms. VRL provides two functions for using enrichment tables:


A single unit of data that flows through Vector. You can learn more about events on the Data model page.


A type of transform that filters events or fields on an event.


The act of sending a batched payload to a downstream service. This is commonly used in conjunction with buffering.


The service used to host Vector’s source code.


A tutorial or walkthrough of a specific subject. You can see Vector’s guides in the Guides section.


An individual log event. Logs are one of the core Vector event types, alongside metrics.


An individual data unit representing a point-in-time measurement. Metrics are one of the core Vector event types, alongside logs.


A transform that parses event data.


The end result of combining sources, transforms, and sinks.


A transform that reduces data into a metric.


A Git repository, usually the Vector Git repository.


A capacity in which Vector is deployed. For more, see the listing of available roles for Vector.


Something that accepts and routes data to many destinations. Vector is commonly referred to as a router.


The Rust programming language. Vector is written exclusively in Rust and takes heavy advantage of Rust’s core features, such as memory efficiency and memory safety.


A transform that samples data (i.e. retains only a subset of a stream based on inclusion criteria).


One of the core component types in Vector, sinks deliver observability to a variety of available destinations. Sinks are the terminus points for Vector pipelines.


One of the core component types in Vector, sources take in observability data from a wide range of available targets. Sources are the entry points for Vector pipelines.

Structured log

A log represented in a structured form, such as a map. Structured logs are distinguished from text logs, which are represented as a single text string.


The TOML table type, which is a collection of key/value pairs directly akin to data types like dicts (Python), objects (JavaScript), and hashes (Ruby).


A deployment topology under which Vector is deployed.


One of the core component types in Vector, transforms perform some kind of action upon events flowing through Vector pipelines, such as sampling, filtering, or modifying events.

Vector Remap Language

An expression-oriented domain-specific language (DSL) that you can use to modify observability data in Vector and also for other tasks, such as specifying Boolean conditions for filtering and routing events. Also known as VRL for short.

For more information, see the remap transform, the announcement blog post, the VRL overview, the comprehensive listing of VRL functions, errors, and examples, and the overview of VRL expressions.