/ compare / loki
[ COMPARE · vs LABEL-FIRST ]
Rootprint
vs
Grafana Loki
rootprint · self-hosted vs loki · label-index
Loki indexes labels; Rootprint indexes full text. Where that distinction matters during an incident — and where Loki's tradeoff still wins.
Last updated 2026-05-29 · written by the rootprint team
TL;DR — COST
Both are object-storage-backed. Costs are close; query shape is what differs.
TL;DR — DATA
Self-hosted on both sides. Rootprint exposes a Quickwit-backed full-text index.
TL;DR — TRADEOFF
Loki rewards disciplined label design. Rootprint trades that for fewer cardinality footguns at the cost of a slightly heavier index.
[ MATRIX ]
Feature-by-feature
A direct comparison on the surfaces most teams care about when picking a logging platform.
| Capability | Rootprint | Grafana Loki |
|---|---|---|
| Hosting model | YESSelf-host on your infra | Self-host (Grafana stack) |
| License | Apache-2.0Open source | AGPL-3.0Open source |
| Ingest path | OTLP/HTTP · NDJSON · Vector · Fluent Bit | Promtail · OTLP · Fluent Bit |
| Index & query | Full-text + structured fields | LABELSLabel filters + LogQL |
| Storage | Your S3 / R2 / MinIO bucket | Object storage (S3 / GCS) |
| Retention pricing | Object-storage rates ($/GB/month) | Object-storage rates |
| Operational surface | Compose stack + object storage | MULTIDistributor / ingester / querier |
| Beyond logs (APM/RUM) | NOLogs only | NOLogs only (Grafana for rest) |
| Data sovereignty | Stays inside your VPC | Stays inside your infra |
| Lock-in risk | NDJSON exports, readable segments | Chunk format + LogQL coupling |
PICK ROOTPRINT IF…
You'd rather own the storage line item.
- ●Long retention is the budget killer.
- ●Compliance / sovereignty requires logs stay in your perimeter.
- ●You already operate object storage and OpenTelemetry collectors.
- ●You want a stack you can read, audit, and fork.
STAY WITH LOKI IF…
You've invested in disciplined labels.
- ○Your teams already design low-cardinality labels well.
- ○You live inside Grafana dashboards already.
- ○Your queries are label-scoped, not full-text scans.
- ○You want the lightest possible index footprint.
[ NEXT ]
Try Rootprint side-by-side with your current pipeline.
Dual-ship for a week and compare the storage line item for yourself.