vectordotdev/vector

Vector is an observability data pipeline, a piece of plumbing that sits between the systems that produce monitoring data (logs and metrics from your applications, servers, and cloud services) and the tools where you want to view, store, or analyse that data. Its job is to collect the data from many sources, transform or filter it along the way, and route it to whichever destinations you choose. Operations teams typically run several different agents on every server, each shipping data to a specific vendor or backend. Vector replaces that patchwork with a single tool that handles logs and metrics in one unified pipeline. You can deploy it on each machine as an agent that gathers data locally, as an aggregator that receives data from many agents, or both. Because the same tool collects, transforms, and forwards data, you can change observability vendors or split a stream across multiple destinations without rewriting the agents on every server. The README highlights a few practical use cases: reducing observability costs (for example by dropping or sampling noisy data before paying a vendor to ingest it), enriching events with extra context, consolidating multiple agents into one, and improving overall pipeline reliability. Performance benchmark tables in the README compare Vector to alternatives such as Filebeat, FluentBit, FluentD, Logstash, and Splunk's forwarders on throughput and correctness tests. The project is written in Rust, which the README cites as the basis of its reliability and speed goals. It is open source and is maintained by Datadog's Community Open Source Engineering team. The full README is longer than what was provided.