High-cardinality metrics stored natively on Apache Pinot. No per-host surcharge, no per-series billing. Cardinality Analytics catches the misconfigured tag before it hits the bill. PromQL-compatible from day one.
The Kloudfuse metrics UI · native dashboards, no Grafana required
No per-host surcharge. Every label, every dimension, stored natively. Predictable pricing on one axis: terabytes ingested.
See which labels, metrics, and services are driving series counts. Catch the unbounded tag at ingest, not on the invoice.
Bring existing Prometheus and Grafana queries unchanged. Reach for the algorithmic functions when a static threshold can't express the question.
Ship from whatever you already run. Kloudfuse normalizes every metric source into one Pinot-backed store. No re-instrumentation, no agent swap.
Cardinality Analytics breaks down series counts by label, metric, namespace, and service. Identify exactly which tag is driving the explosion, and bound it at ingest — not after the invoice.
Cardinality Analytics — series counts broken down by metric, label and label value
Fully compliant PromQL interface. Existing Prometheus and Grafana queries work without changes. Import your Grafana dashboards directly, or start with pre-built dashboards for Kubernetes, infrastructure, and common services. Kloudfuse extends PromQL with algorithmic functions for anomaly detection and forecasting, so those queries run in the language your dashboards already use.
Metrics explorer — a PromQL query charted with its label breakdown
Multi-rollup computes aggregations at ingest in configurable intervals. Long-range queries hit rollups. Drill-down hits raw. For most common use cases, multi-rollup handles what recording rules used to. For complex queries like custom SLOs, recording rules are still available and recommended.
| Resolution | Granularity | Serves | Selected |
|---|---|---|---|
| Raw | Every point · 15s | Drill-down, incident forensics | Automatically |
| Rollup · 5-minute | Pre-aggregated | Day- and week-range dashboards | Automatically |
| Rollup · 1-hour | Pre-aggregated | Long-range trends and capacity planning | Automatically |
Metrics live in your VPC, in Pinot running on your cloud account. The bill scales with what your platform runs, not with how many tags your team adds.
| Billing axis | Typical SaaS | Kloudfuse |
|---|---|---|
| Per-host surcharge | Yes | None |
| Per-series billing | Yes | None |
| Custom metrics tax | Yes | None |
| Pricing model | Six independent axes | One axis · predictable |
| Retention | Priced tiers | Tier-aware · your control |
No. Every dimension is stored natively. There is no premium tier, no per-custom-metric surcharge. Pricing runs on one axis: terabytes ingested.
A built-in view that breaks down series cardinality by label, metric, namespace, and service. It surfaces the specific tags driving cardinality spikes so engineers can bound them at ingest.
Yes. The PromQL interface is fully compliant. Existing Prometheus and Grafana workflows work without modification. Kloudfuse extends PromQL with algorithmic functions for anomaly detection and forecasting, so those queries run in the language your dashboards already use.
Yes. Kloudfuse ships with pre-built Grafana-compatible dashboards and alerts for Kubernetes monitoring, infrastructure health, and common services. You can also import any existing Grafana dashboard — they work without modification.
Yes. Multi-rollup produces pre-aggregated data in configurable intervals alongside raw metric retention. Long-range queries hit rollups automatically. For most common aggregations, this drastically reduces or eliminates the need for recording rules. Recording rules remain available and recommended for complex or expensive queries like custom SLOs.
Inside your VPC, in Apache Pinot running on your cloud account, under your keys.
See which labels are driving your series count, drill from spike to source tag, and set bounds before cardinality hits the bill.
Read the guideHow cardinality management, configurable rollups, and one-axis pricing change the economics of metrics at scale.
Read the blogPricing model, cardinality handling, data residency, agent compatibility. A side-by-side breakdown for teams evaluating both.
See the comparisonThirty minutes on your telemetry. The cause, before the call ends.