● Beta — KubeEra is in active development. Profiles are live; AI-perception measurement is rolling out.
KubeEra
V

Vector

Observability
Open source · MPL-2.0

Vector is a high-performance observability data pipeline. It collects, transforms, and routes logs, metrics, and traces from wherever they originate (files, syslog, Kafka, StatsD, cloud provider logs, etc.

updated 2026-07-01

What it is

Vector is a high-performance observability data pipeline. It collects, transforms, and routes logs, metrics, and traces from wherever they originate (files, syslog, Kafka, StatsD, cloud provider logs, etc.) to wherever they need to go (Elasticsearch, S3, Datadog, Splunk, Kafka, ClickHouse, and dozens more). Think of it as a replacement for Logstash, Fluentd, or Telegraf — but written in Rust for lower memory footprint and higher throughput per core.

It runs as either an agent (on every host/pod, collecting local data) or an aggregator (centralized, receiving from many agents). Configuration is declarative (TOML/YAML/JSON), with a VRL (Vector Remap Language) for transforms — a purpose-built DSL instead of embedding Lua or full Ruby like some competitors.

Who builds it and why

Vector was originally built by Timber.io, acquired by Datadog in 2021. Datadog continues to fund core development and uses Vector internally as part of its own data pipeline infrastructure, but Vector remains independently open source under MPL-2.0 — it is not gated behind a Datadog subscription, and it ships first-class sinks for competing platforms (Splunk, Elastic, New Relic, Grafana Cloud) alongside Datadog's own.

The incentive alignment is reasonably clear: Datadog wants a best-in-class, vendor-neutral collection layer because it lowers the cost of switching data into Datadog, and because a widely-adopted open pipeline benefits their own ingestion economics. Whether that alignment holds indefinitely is a fair question to ask before betting critical infrastructure on it.

Production readiness signal

22,113 GitHub stars and 634 contributors is a real, active project — not a side experiment. Commits as recent as July 2026 indicate ongoing maintenance rather than abandonment. CNCF maturity level is not publicly available, meaning Vector is not (or not yet) part of the CNCF landscape in a formal sandbox/incubating/graduated sense — evaluate it on its own engineering merits, not a foundation stamp.

The release naming (vdev-v0.3.4) is worth noting: this looks like tooling/dev-environment versioning rather than the main Vector product release train, which historically uses semantic versions like 0.4x.x. Confirm you're tracking the actual product release cadence, not internal dev tooling, before making assumptions about stability from this tag alone.

Rust gives you memory safety and performance without a GC pause tax — a real advantage over JVM-based pipelines under sustained high-cardinality load.

Who should use this

  • Teams running multi-destination observability pipelines (e.g., logs to both S3 for archive and Elasticsearch for search) who are tired of maintaining multiple single-purpose shippers.
  • Anyone hitting memory/CPU ceilings with Fluentd or Logstash at high log volume per node.
  • Platform teams that want a single agent binary across logs, metrics, and traces instead of three separate tools.
  • Organizations wanting to decouple from a single observability vendor — Vector's sink-agnostic design is genuinely useful leverage in vendor negotiations.

Who should NOT use this

  • Small shops with low log volume and a single destination — Fluent Bit is lighter-weight and simpler to reason about for that case.
  • Teams needing a mature plugin ecosystem for obscure legacy log formats — Logstash's 15+ years of community plugins still cover more edge cases than Vector's growing-but-younger sink/source library.
  • Anyone uncomfortable maintaining VRL transform logic — it's powerful but is another DSL your team has to learn and debug in production incidents.
  • Shops wanting a fully managed, zero-ops collection layer — Vector is self-hosted infrastructure; you own the upgrade cycle, config drift, and capacity planning.

Alternatives

  • Fluent Bit / Fluentd — larger plugin ecosystem, CNCF graduated project, lower resource use for simple cases but generally lower throughput ceiling than Vector.
  • Logstash — most mature plugin library, but JVM overhead makes it heavier per unit of data processed.
  • OpenTelemetry Collector — better native fit if you're standardizing on OTel for traces/metrics specifically, though its log pipeline maturity still trails Vector's.

Pricing

Fully open source under MPL-2.0. No paid tier, no license gating of features. You pay only in infrastructure and engineering time to run it yourself.

Frequently asked

What is Vector?+
Vector is a high-performance observability data pipeline. It collects, transforms, and routes logs, metrics, and traces from wherever they originate (files, syslog, Kafka, StatsD, cloud provider logs, etc.) to wherever they need to go (Elasticsearch, S3, Datadog, Splunk, Kafka, ClickHouse, and dozens more).
Who builds Vector?+
Vector was originally built by Timber.io, acquired by Datadog in 2021. Datadog continues to fund core development and uses Vector internally as part of its own data pipeline infrastructure, but Vector remains independently open source under MPL-2.
Is Vector production ready?+
22,113 GitHub stars and 634 contributors is a real, active project — not a side experiment. Commits as recent as July 2026 indicate ongoing maintenance rather than abandonment.
Who should use Vector?+
Teams running multi-destination observability pipelines (e.g., logs to both S3 for archive and Elasticsearch for search) who are tired of maintaining multiple single-purpose shippers. Anyone hitting memory/CPU ceilings with Fluentd or Logstash at high log volume per node.
Who should not use Vector?+
Small shops with low log volume and a single destination — Fluent Bit is lighter-weight and simpler to reason about for that case.
What are the alternatives to Vector?+
Fluent Bit / Fluentd — larger plugin ecosystem, CNCF graduated project, lower resource use for simple cases but generally lower throughput ceiling than Vector. Logstash — most mature plugin library, but JVM overhead makes it heavier per unit of data processed.
How much does Vector cost?+
Fully open source under MPL-2.0. No paid tier, no license gating of features. You pay only in infrastructure and engineering time to run it yourself.