Real-time
Metrics.
Every counter, gauge, and histogram — live. Unified dashboards that update in under a millisecond, with ML-powered anomaly detection built in.
<1ms
Metric refresh rate
∞
Cardinality limit
13mo
Full-res retention
400+
Native integrations
What are real-time metrics?
Numbers that reflect your system right now — not a minute ago.
Traditional monitoring tools poll at 15- or 60-second intervals. By the time a spike appears on your dashboard, it may already have caused customer impact. obseria.io's metrics pipeline uses a streaming architecture to process, aggregate, and deliver gauge readings, counters, and histograms with sub-millisecond end-to-end latency.
Every metric is enriched with dimensional labels — service, environment, region, Kubernetes pod, and any custom attribute you add — without any cardinality limits or extra cost. Query across billions of time-series data points with PromQL or obseria.io's native DSL, and build dashboards that update in real time as you type.
Requests/sec
1,847
+3.2%
p95 Latency
84ms
−6ms
Error Rate
0.21%
+0.03%
Apdex Score
0.97
+0.02
Requests per second
CPU · web
CPU · api
CPU · db
Memory
Multi-dimensional data
Break down any metric by any dimension.
Group by service, region, pod, release version, or any custom label — no pre-aggregation required.
CPU Utilization
by service · last 48 hours
Error Rate
all services · last 30 days
How it works
From instrument to insight in three steps.
obseria.io's metrics pipeline is OpenTelemetry-native — no proprietary agents required.
01
Emit metrics via OTLP or Prometheus
Send counters, gauges, and histograms via OpenTelemetry SDK, Prometheus remote-write, StatsD, or any of 400+ native integrations. No agent to install — a single endpoint handles everything.
02
Stream processing with no pre-aggregation
obseria.io's ingest layer processes every data point individually. Aggregate on query — not at ingest time — so you never lose granularity and can answer questions you haven't thought of yet.
03
Live dashboards + AI alerting
Dashboards reflect your system state in under 1 ms. ML baselines automatically detect anomalies relative to your traffic patterns — no static thresholds to maintain or tune.
Capabilities
Everything your on-call team needs — and then some.
Built for production-scale engineering teams, not demo environments.
Sub-millisecond refresh
Metrics update in under 1 ms. No polling lag — changes in your system appear on the dashboard before users notice.
Smart anomaly alerts
ML baselines learn your traffic patterns over 14 days. Alerts fire on true anomalies, not arbitrary static thresholds.
Unlimited cardinality
Filter by any combination of service, region, pod, customer ID, or custom label — no cardinality limits.
Multi-source federation
Scrape Prometheus, StatsD, OpenTelemetry, and cloud-native metrics side-by-side in a single namespace.
Forecasting & capacity
30-day trend models predict when you'll hit CPU, memory, or throughput limits — plan infra before incidents happen.
Rollback correlation
Deployments are overlaid automatically on every chart. See exactly which release caused a metric to degrade.
13-month retention
Full-resolution metric data stored for 13 months — useful for year-over-year capacity reviews and SLA audits.
Metric Explorer
Ad-hoc query builder with PromQL and obseria.io's own DSL. Export results to CSV, S3, or your data warehouse.
Custom dashboards
Drag-and-drop builder with 30+ widget types. Embed dashboards in Confluence, Notion, or any iframe target.
Customer stories
Engineering teams that switched from Datadog and Grafana.
“We were paying Datadog $180k/year and still hitting cardinality limits. obseria.io costs us $28k annually with no limits. The dashboards are faster and the anomaly alerts actually fire on real problems.”
Tom Lindqvist
Director of Platform Engineering · Solvra
“The <1ms refresh is not a marketing claim — I can watch individual request spikes disappear in real time during load tests. Our on-call team now catches CPU saturation before the auto-scaler even triggers.”
Ava Moreau
Lead SRE · Crescent Systems
“We migrated 2,400 Grafana dashboards in one weekend using the import tool. The ML baselines had our services modeled within three days and immediately started firing useful alerts we hadn't configured before.”
James Park
VP Infrastructure · Loopscale
Trusted by 3,400+ engineering teams
SolvraCrescent SystemsLoopscaleAscend HealthMeridian CloudStackfireFAQ
Common questions about real-time metrics.
How does obseria.io achieve sub-millisecond metric refresh?
obseria.io uses a streaming ingest pipeline built on Apache Kafka and a custom columnar store optimised for time-series fan-out. Metrics are never batched before delivery to the query layer — each data point flows directly to the live dashboard WebSocket connection.
Can I use PromQL to query my metrics?
Yes — PromQL is fully supported as a first-class query language. You can also use obseria.io's native DSL for cross-signal queries that join traces, logs, and metrics in a single expression. Both languages work in dashboards, alerts, and the Metric Explorer.
What is unlimited cardinality and why does it matter?
Cardinality refers to the number of unique label combinations on a metric. Most tools charge extra or drop data when cardinality is high. obseria.io charges per data point, not per unique series — so you can safely add high-cardinality labels like customer_id or request_path without budget surprises.
How does the ML anomaly detection work?
obseria.io models the baseline distribution of each metric series over a 14-day rolling window. When an incoming value falls outside the expected range — accounting for daily and weekly seasonality — an alert fires automatically. There are no thresholds to configure; the model adapts as your traffic patterns change.
Can I migrate my existing Grafana dashboards?
Yes. obseria.io's migration tool imports Grafana JSON exports and translates panel configurations automatically. PromQL queries are compatible without modification. Most customers complete migration of hundreds of dashboards in a single weekend.
How is obseria.io pricing calculated for metrics?
You pay per million data points ingested — currently $0.18/M DPM. There are no per-series, per-dashboard, or per-user charges. Budget caps prevent unexpected bills, and the usage dashboard shows your projected monthly cost in real time.
Enterprise
Need custom retention, SLA, or volume pricing?
For large-scale infrastructure teams we offer dedicated ingest clusters, 13-month retention, HIPAA BAA, SAML/SCIM provisioning, and a dedicated Customer Success Manager.
- Volume discounts from 50B DPM/month
- Custom retention up to 13 months
- Dedicated ingest cluster
- Grafana migration assistance
- SOC 2 Type II + HIPAA BAA
- 24/7 phone SLA + dedicated CSM
Get in touch
Typically respond within 1 business hour.
See your metrics live in minutes.
Point your existing Prometheus or OpenTelemetry setup at obseria.io's OTLP endpoint. Dashboards appear automatically — no configuration required.