Vector

Vector is an open source, high-performance observability

data pipeline that collects, transforms, and routes logs, metrics, and traces (observability, data pipeline) across heterogeneous infrastructure.

• Ingests logs, metrics, and traces from multiple sources and agents into a unified pipeline (observability data collection).
• Applies transformations, filtering, enrichment, and routing to telemetry data via a configurable processing graph (data processing and Extract, Transform, Load (ETL)).
• Outputs data to various backends and vendors, including observability platforms and storage systems (data routing and integration).
• Provides a single, vendor-agnostic pipeline that can consolidate agents and reduce telemetry overhead (telemetry pipeline management).
• Supports deployment across containers, virtual machines, bare metal, and cloud environments for centralized observability data handling (infrastructure observability).

More About Vector

Vector is a programmable, open source observability data pipeline designed to collect, transform, and route telemetry data, including logs, metrics, and traces (observability, data pipeline). It sits between application and infrastructure sources and downstream observability or storage systems, giving enterprises a single control point for how telemetry flows through their environments. Vector focuses on providing a unified pipeline that can standardize data processing and routing logic across diverse platforms and vendors.

The project exposes a modular architecture based on sources, transforms, and sinks (pipeline orchestration). Sources handle ingestion from agents, services, and infrastructure components, capturing log streams, metrics, and distributed traces. Transforms provide capabilities for parsing, filtering, aggregating, enriching, remapping, and shaping telemetry records, allowing teams to normalize data into consistent schemas or reduce volume before it reaches downstream systems. Sinks define destinations such as observability platforms, object stores, message queues, and analytics systems, enabling routing policies that can duplicate, split, or selectively forward data to multiple backends.

Vector is built to run in varied enterprise deployment models (infrastructure observability). It can operate as a sidecar or daemonset in container orchestration environments, as an agent on virtual machines or bare metal hosts, or as an intermediate service within centralized logging and metrics architectures. Configuration is typically expressed declaratively, enabling repeatable deployments through Infrastructure-as-Code (IaC) workflows and integration with configuration management tools.

The project’s processing graph supports backpressure-aware, streaming data handling (stream processing). Vector emphasizes performance and resource efficiency in order to handle high-throughput telemetry pipelines. It supports structured data formats, common logging encodings, and metrics protocols depending on configured sources and sinks, and can bridge formats between producers and consumers.

For interoperability, Vector positions itself as a vendor-agnostic pipeline that can send data to multiple observability platforms and storage technologies (toolchain integration). This allows organizations to standardize on one collection and processing layer while retaining flexibility in downstream analytics and monitoring tools. Extensibility is provided through a growing catalog of components and configuration-driven behaviors rather than custom code changes in most cases.

Within an enterprise taxonomy, Vector fits into observability pipelines, log management, metrics and traces forwarding, and data engineering for telemetry (observability, data engineering). It is used to centralize control over data quality, sampling, routing, and cost management for logs and metrics, and to reduce the number of agents or bespoke collectors that must be operated and maintained across large-scale infrastructures.