Grafana Launches Adaptive Logs Drop Rules to Cut Noise and Costs in Public Preview

Grafana Introduces Adaptive Logs Drop Rules in Public Preview

Grafana Labs today announced a new feature for its Adaptive Logs product: customizable drop rules that let teams eliminate noisy, low-value log lines before they are ingested into Grafana Cloud Logs. The capability, now in public preview, promises immediate cost savings and reduced operational noise for platform and observability teams.

Grafana Launches Adaptive Logs Drop Rules to Cut Noise and Costs in Public Preview

“Most organizations have logs they know are noise—health checks, forgotten DEBUG statements, verbose INFO from rarely used services. Until now, there hasn’t been a simple way to drop them without complex infrastructure changes,” said Alex Turner, Senior Product Manager at Grafana Labs. “With drop rules, teams can define their own logic to filter out that waste instantly.”

Key Details

Drop rules allow users to specify criteria based on log labels, detected log levels, or line content. Logs matching a rule are dropped before being written to Cloud storage, reducing both noise and costs. This feature complements the existing intelligent optimization recommendations in Adaptive Logs, already available for Adaptive Metrics and Adaptive Traces.

How Drop Rules Work

Each drop rule applies a configurable drop rate (0–100%). Users can target specific services, log levels, or even repetitive line patterns. For example:

Drop all DEBUG logs from a development environment.
Sample chatty but necessary logs by applying a 90% drop rate.
Isolate a single misbehaving service emitting high-volume, low-value logs.

Turner emphasized the simplicity: “A platform team can create one rule with a 100% drop rate for health check logs, and every service benefits without any code changes.”

Integration with Adaptive Logs

Drop rules are part of a three-step processing pipeline in Adaptive Logs. When a log line arrives:

Exemptions protect critical logs from any sampling.
Drop rules are evaluated in priority order; the first matching rule applies its drop rate.
Patterns (optimization recommendations) handle the remaining lines.

This layered approach ensures that only truly noisy logs are removed, while valuable data is preserved. “Drop rules give teams direct control to eliminate known noise, supplementing our automated recommendations,” Turner added.

Background

Centralized observability teams have long struggled with noisy logs that inflate bills and obscure real issues. Traditional solutions required changing application code or managing infrastructure configuration—both slow and error-prone. Grafana Cloud already offered similar custom drop capabilities for metrics and traces, but logs remained a challenge until now.

“We heard from customers that log noise was their number one cost concern,” said Turner. “Adaptive Logs drop rules close that gap, giving them the same control they already have for metrics and traces.”

What This Means

For platform and observability teams, the immediate benefit is lower costs and cleaner dashboards. By dropping known noise (health checks, DEBUG logs, repetitive batch processing logs) before ingestion, organizations can reduce log volume by up to 50% or more without losing signal. The feature also reduces cognitive load on engineers who no longer need to sift through irrelevant log lines.

“This isn’t just about saving money—it’s about improving the reliability of your observability data,” Turner noted. “When you remove noise, alerts become more meaningful and troubleshooting faster.” The feature is available starting today for all Grafana Cloud customers as a public preview, with full documentation on the Grafana Labs website.

For details on creating drop rules, see the Adaptive Logs documentation.