The challenge
Stop paying platform prices for data nobody queries
Observability platforms bill by log volume, metric series or data points, and trace volume or host count. Much of it never reaches a dashboard, an alert or a query, and the bill grows as you add services.
The Solution
Control what reaches your observability platform
Cut volume without losing what you troubleshoot with
Set reduction rules for logs, metrics and traces before data reaches your platform. Drop repeats and noise, sample high-volume streams, and keep the data you troubleshoot with. Start from ready-made rules for OpenTelemetry, Prometheus, the Datadog Agent and Kubernetes, or write your own.
Send each copy where it costs least
Send the reduced stream to your observability platform and a full copy to storage you own. When an incident needs more, replay a time window from storage back through your pipelines.
Measure savings at every step
Preview what a rule removes on your own sample data before it goes live. In production, see how much data comes in and how much reaches each platform, for every route. Track the reduction in the dashboards you already use.
Reduce without risk
Every reduction is a rule you set, test against sample data, and promote from staging to production. Each destination has its own rules, so cuts made for your observability platform don’t change what your SIEM receives.
Key benefits
Why this approach works
ROI & SAVINGS
Measurable impact on your observability budget
DataStream sends less telemetry to your platform and handles more of it on the same hardware. Throughput comes from PipeBench, our open benchmark.
Telemetry sent to your platform
Up to 60% less telemetry in our tests, from filtering logs, sampling traces and merging metric series.
OTLP/HTTP events per second on a 16-vCPU host
Nearly 10 times the throughput on the same host, so the pipeline itself needs fewer hosts.
Frequently asked questions
Can we turn high-volume logs into metrics?
Yes. DataStream rolls logs up into counts, sums, minimums, maximums and averages per time window, grouped by any field. It sends them as metrics to platforms that take metrics. Each platform gets the temporality it expects: cumulative counters for Prometheus remote write, deltas for Dynatrace or Datadog.
Do you drop data by default?
No. DataStream drops nothing until you add a rule. Each destination still follows its platform’s limits: a platform that takes no logs gets none.
How do you reduce metric cardinality?
Remove the metric labels you never query and merge the series they leave behind over a time window, or cap the number of series per metric. This works on metrics from OpenTelemetry, Prometheus, the Datadog Agent and StatsD.
Can we mask sensitive data before it reaches our platform?
Yes. Pipeline steps mask or redact sensitive values, such as email addresses, IP addresses and card numbers, before data reaches any destination.
Talk to our experts
Schedule a technical session with our team to review your observability costs and see how DataStream reduces them safely.
Try DataStream
Try DataStream in your environment and measure the impact on your observability ingestion costs in weeks.
Start now