The Vector blog on Vector

Highlights - July 2026

Thu, 02 Jul 2026 00:00:00 +0000

It has been a while since our February 2025 highlights. In this post we cover the most impactful features shipped across 0.46 through 0.56, and a set of CI improvements that make life easier for external contributors.

Features

New components

Component	Kind	Description	Author	PR
`mqtt`	Source	Ingest from MQTT brokers	@StormStake	#22752
`okta`	Source	Consume Okta system logs	@sonnens	#22968
`websocket`	Source	Real-time ingestion from WebSocket APIs	@benjamin-awd	#23449
`windows_event_log`	Source	Native Windows Event Log with bookmark-based checkpointing	@tot19	#24305
`delay`	Transform	Delay events by a fixed duration or VRL condition	@esensar, @Quad9DNS	#25407
`incremental_to_absolute`	Transform	Reconstruct absolute metrics from incremental data	@GreyLilac09	#23374
`trace_to_log`	Transform	Convert traces to logs	@spencerho777	#24168
`window`	Transform	Sliding-window ring-buffer for noise reduction	@ilinas	#22609
`azure_logs_ingestion`	Sink	Send logs to Azure Monitor via the Logs Ingestion API	@jlaundry	#22912
`databricks_zerobus`	Sink	Stream to Databricks Unity Catalog via Zerobus	@flaviofcruz	#24840
`doris`	Sink	Apache Doris via the Stream Load API	@bingquanzhao	#23117
`postgres`	Sink	Send logs, metrics, and traces to Postgres	@jorgehermo9	#21248
`otlp`	Codec	Bidirectional Vector <-> OTLP conversion (logs and traces)	@pront	#24003
`syslog`	Encoder	Encode Vector events as syslog (RFC5424 and RFC3164)	@vparfonov	#23777
`varint_length_delimited`	Framer	Varint length-delimited framing for protobuf streaming (ClickHouse-compatible)	@modev2301	#23352

Source improvements

The opentelemetry source gained metrics ingestion and now performs full OTLP decoding for logs, metrics, and traces, removing the need for complex remap steps in OTEL -> Vector -> OTEL pipelines.
The docker_logs source retries Docker daemon communication failures with exponential backoff instead of giving up on transient hiccups.
A performance regression that inflated CPU usage in the file and kubernetes_logs sources (introduced in 0.50.0) was found and fixed.

Transform improvements

The tag_cardinality_limit transform gained several new controls: per-tag cardinality overrides (per_tag_limits), per-metric tracking isolation (tracking_scope: per_metric), a global key cap (max_tracked_keys), and the ability to opt entire metrics out of cardinality tracking.
The syslog encoding transform gained improved RFC compliance, support for scalars, nested objects, and arrays in structured data, and better UTF-8 safety.

Sink improvements

A configurable retry_strategy for HTTP-based sinks gives users control over which response codes are retried (default / none / all / custom).
The aws_s3 sink gained Apache Parquet batch encoding.
The datadog_metrics sink switched to the Series v2 endpoint with zstd compression by default, and datadog_logs switched its default to zstd as well.

Operations and observability

--watch-config now also watches enrichment table files.
vector top gained scrollable, sortable, and filterable views (press ? for keybinds).
Unit tests support an expected_event_count field on outputs to assert on emitted event counts.
Task-transform utilization no longer counts downstream wait time, giving a more accurate saturation view.
New internal metrics for capacity planning and backpressure detection:
- source_buffer_max_size_bytes, source_buffer_max_size_events
- transform_buffer_max_size_bytes, transform_buffer_max_size_events
- source_buffer_utilization_mean, transform_buffer_utilization_mean (EWMA-smoothed)
- component_latency_seconds (histogram), component_latency_mean_seconds (gauge)
- source_send_latency_seconds, source_send_batch_latency_seconds

For the complete list of changes, breaking changes, and upgrade steps, see the releases page.

Celebrating COSE's First Year

Thu, 13 Nov 2025 00:00:00 +0000

Our journey

In October 2024, the COSE (Community Open Source Engineering) team was formed with the mission to strengthen Vector’s open source foundation and improve the developer experience. Today, we’re celebrating our first year by highlighting the contributions we’ve made to the Vector community. The COSE team has committed over 550 changes to Vector, across 8 major releases (0.43.0 through 0.51.0).

Highlights - February 2025

Wed, 12 Feb 2025 00:00:00 +0000

In this blog post we want to highlight some recently added features and guides.

Features

A new OpenTelemetry sink is now available. You can find a quickstart guide here.
- We have more plans here such as gRPC support and smart grouping at the sink.
We introduced a new exclusive route transform. You can read more in our release highlight.
Apple Silicon builds are now available in our downloads page. Note that these are available since v0.44.0 and won’t show up for older versions.
A new type of enrichment_table, called memory, was introduced! This table can also function as a sink and addresses several new use cases. See this release highlight for details.
The VRL function library keeps growing thanks to community contributions! You can read more:
- 0.21.0 Features
- 0.20.0 Features

Guides

We recently added a new guide category: Development.

Log Namespacing

Fri, 30 Jun 2023 00:00:00 +0000

Please visit our log namespace guide for more details.

The Vector team has been hard at work improving the data model of events in Vector. These changes are now available for beta testing for those who want to try it out and give feedback. This is an opt-in feature. Nothing should change unless you specifically enable it.

Monitoring Vector's per-component memory usage

Wed, 14 Dec 2022 00:00:00 +0000

We are excited to announce that Vector now has support for exposing per-component memory usage metrics. This work addresses an often-requested feature from users who want to understand how Vector uses memory, and what parts of their configurations are responsible for high memory usage.

Trying it out

Vector Remap Language

Mon, 15 Feb 2021 00:00:00 +0000

Vector Remap Language (VRL) is an expression-oriented language designed to work with observability data (logs and metrics) in a safe and performant manner. It features a simple syntax, a rich set of built-in functions tailored to observability use cases, and numerous features that set it far apart from other options. This 0.12 release of Vector marks the official release of the language.

Accelerating Vector with Datadog

Thu, 11 Feb 2021 00:00:00 +0000

We’re excited to announce that Timber and Vector are now officially part of Datadog!

When we started building Vector, we set out to provide an entirely new kind of data pipeline. We were guided by a belief that when it comes to collecting, transforming, and routing observability data, teams shouldn’t have to accept trade-offs between cost, performance, reliability, and security. They also shouldn’t have to stitch together disparate tools just to get their data from point A to B; the ideal solution should be simple, yet robust, and unify the entire pipeline. We’ve dedicated ourselves to that challenge, and we’re proud of the platform we’ve created.

The new GraphQL API for Vector

Tue, 08 Dec 2020 00:00:00 +0000

The Vector observability API migrated from GraphQL to gRPC in version 0.55.0.

Although Vector is an observability tool, it’s nonetheless important to be able to observe Vector itself, especially in production environments where it serves as critical infrastructure. That’s why we’re excited to announce the new Vector GraphQL API, available in v0.11.0.

First-class Kubernetes integration for Vector

Thu, 26 Nov 2020 00:00:00 +0000

After months of development, over 100 pull requests, and intensive QA in clusters producing over 20 terabytes of event data a day, we’re pleased to announce our first-class Kubernetes integration for Vector in version 0.11. We strove to make even this initial integration rock solid and production ready because we aim to make Vector the default pipeline for all Kubernetes observability data.

To cut straight to the chase, checkout the installation instructions, otherwise read on for the details.

Adaptive request concurrency. Resilient observability at scale.

Tue, 24 Nov 2020 00:00:00 +0000

Observability pipelines have become critical infrastructure in the current technological landscape, which is why we’ve built Vector to provide extremely high throughput with the tiniest resource footprint we can manage (Rust is a huge help here). But this is not enough in the real world: your observability pipeline needs to provide optimal performance and efficiency while also being a good infrastructure citizen and playing nicely with services like Elasticsearch and Clickhouse.

And so we’re excited to announce that Vector version 0.11 includes support for Adaptive Request Concurrency (ARC) in all of its HTTP-based sinks. This feature does away with static rate limits and automatically optimizes HTTP concurrency limits based on downstream service responses. The underlying mechanism is a feedback loop inspired by TCP congestion control algorithms.

How we test Vector

Mon, 13 Jul 2020 00:00:00 +0000

When we set out to build Vector, we knew that reliability and performance were two of our top priorities. We also knew that even the best of intentions would not be enough to make certain those qualities were realized and reflected in our users’ production deployments. Since then, we’ve been continuously evolving and expanding our approach to achieving that level of quality.

Introducing Vector

Mon, 13 Jul 2020 00:00:00 +0000

Today we’re very excited to open source the Vector project! Vector is a tool for building flexible and robust pipelines for your logs and metrics data. We’re still in the early stages, but our goal with Vector is to dramatically simplify your observability infrastructure while making it easy to get more value from your data.