Grafana Loki 3.4: Standardized storage config, sizing guidance, and Promtail merging into Alloy
The Grafana Loki 3.4 release is here, and it brings a fresh wave of enhancements aimed at standardizing Loki’s object storage, helping you right size your instance, and improving the ability to ingest out-of-order logs.
Loki 3.4 also represents the official merging of Promtail into Grafana Alloy as part of our efforts to give our users a single telemetry collector. There’s a lot to go over, so let’s dive in.
Loki feature updates
New Thanos-based client for configuring object storage
Loki 3.4 introduces support for the Thanos Object Storage Client, making storage configuration more consistent across Grafana’s OSS databases. Thanos, a widely adopted CNCF project, provides highly available long-term storage and querying for Prometheus metrics. By adopting its object storage client, Loki is aligning with a proven, well-maintained open source standard that many in the observability community already rely on.
If you’re already using Grafana Mimir or Grafana Pyroscope, you can now reuse the same storage configuration for Loki, reducing setup effort and duplication.
The Thanos object storage config is optional in Loki 3.4, giving you the chance to migrate at your own pace. In a future release, we’ll make the Thanos configuration the default storage client and mark the existing storage configurations as deprecated.
Adopting the Thanos Object Storage Client aligns with our philosophy of embracing open standards, and we want to thank the Thanos maintainers for their work in building and maintaining this great open source library!
New sizing guidance
Loki 3.4 marks the return of the sizing guidance page, which helps users determine the amount of CPU and memory they need to run Loki in microservices mode. We’ve updated it to reflect our latest learnings from running Loki to power Grafana Cloud Logs. Based on the expected ingestion volume, Loki clusters are categorized into three tiers (or sizes) of deployments. Recommendations are based on p90 resource utilization for each relevant component.
Detailed sizing guidance has been a frequent request from customers and users who are trying to figure out how to size their deployments, so we’re excited to make it available in this latest release.
More ways to get value from your logs faster
- Loki’s out-of-order support has been expanded to allow for ingestion of arbitrarily old log lines. To opt in to this feature, set
time_sharding_enabledto true. - It’s now easier to get the benefits of structured metadata with a new feature for extracting fields into structured metadata at ingestion time.
- The Loki Helm chart saw several improvements, including support for
topologySpreadConstraintsand Loki’s overrides exporter.
Promtail merged into Alloy
Over the years, Grafana Labs has developed several different agents to help users collect telemetry data. We built Grafana Agent, launched Promtail for Loki logs, added the Grafana Agent Operator for Kubernetes-specific cases, and more. Each of these has served an important role, but we’ve heard consistent feedback that operators prefer to have one tool that covers all their observability needs.
As a result, we have been working towards a streamlined, unified, and open-standards-based approach to telemetry collection, which resulted in last year’s release of Grafana Alloy, our open source distribution of the OpenTelemetry Collector and the only one containing a high-performance pipeline for native Prometheus metrics.
As we work toward this new vision for telemetry collection, we’re taking the next step by deprecating Promtail and fully merging its capabilities into Alloy.
This move has been a long time in the making, and we’re confident Alloy can maintain Promtail’s functionality without degrading user experience. We’ve also built migration tooling to make the transition as easy as possible, so Loki users can adopt Alloy with minimal overhead.
The Promtail deprecation timeline is as follows:
- 2025-02-13: Long-Term Support (LTS): Promtail enters LTS phase where commercial support, security updates, and critical fixes will continue to be provided.
- 2026-03-02: End-of-Life (EOL): No further support or updates will be provided.
Alloy is the most efficient and cost-effective way to collect Prometheus-compatible metrics. It removes the need to have separate agents or clients to collect different telemetry signals. And unlike Promtail, Alloy is 100% OTLP compatible, with native pipelines for Prometheus and support for metrics, logs, traces, and profiles. If you’re collecting more than just logs, Alloy will reduce the number of agents you need to install, configure, and manage.
Migrating from Promtail to Alloy
We’ve created migration documentation that you can find in both the Loki and Alloy documentation. It includes a migration tool for converting your Promtail configuration to an Alloy configuration with a single command.
If you’re looking for more hands-on ways to learn about Alloy, we’ve got several interactive tutorials:
If you prefer the OpenTelemetry Collector, we’ve got that covered, too:
Read the latest docs to help you get started with Loki 3.4 today
We’re excited for you to try out all the changes in Loki 3.4. We’re also working hard to update the Loki docs so you’re better equipped to get started.
This release includes a heavily revised labels topic and a new cardinality topic. Continuing updates to the Query section include a revised logcli topic and new logcli tutorial. There are also two additional new interactive tutorials: Deploying Loki on Azure tutorial and Kubernetes Monitoring Helm tutorial.
Grafana Cloud is the easiest way to get started with metrics, logs, traces, dashboards, and more. We have a generous forever-free tier and plans for every use case. Sign up for free now!
