Grafana Labs self-hosted stack: Monitoring Ledger 2026
Grafana Labs offers a full self-hosted observability stack with the best release record here, but lacks network management and topology root cause.
Part of the Infrastructure Monitoring Ledger (edition 2026-10-04). How the scores work: methodology.
Snapshot
- Tier: Challenger
- Owner: Grafana Labs (Raintank, Inc.), independent, venture-backed (Lightspeed, Sequoia, GIC); last round USD 270M in 2024
- Licence model: Open core: Grafana, Mimir, Loki, Tempo under AGPLv3 since 2021, Alloy under Apache 2.0; Grafana Enterprise by quote-based subscription
- Products scored: Grafana OSS and Enterprise 13.x, Mimir, Loki, Tempo, Pyroscope and Alloy, all self-hosted; Grafana Cloud excluded
- Latest release: Grafana 13.2.0, 2026-08-18
- Next signal: No dated milestone; direction is generative dashboard authoring, the Agent-to-Alloy migration and Scenes 2.0 across all views
- Archetype: Composable observability stack
- Evidence grade: B. Release dates match git tags; several AI and app features (ML, Sift, Application Observability) are documented for Grafana Cloud and their self-hosted status is unclear.
Scorecard
| Dimension | Score | Direction |
|---|---|---|
| Capability today | 77.8 / 100 | higher is better |
| Momentum since end of 2022 | +10.2 points | higher is better |
| Credibility | 22.0 / 25 | higher is better |
| Commercial risk | 9.5 / 25 | lower is better |
| Technical lock-in | 5.5 / 25 | lower is better |
| Integrator fit | 15.0 / 20 | higher is better |
| AI leverage | 13.0 / 20 | higher is better |
| Ledger Index | 75.4 / 100 | balanced weights |
Capability by domain
Scores 0 to 5 against fixed anchors; total is weighted to 100.
| Domain (weight) | 2016 | 2019 | 2022 | Today |
|---|---|---|---|---|
| Server, virtualization and storage (12) | 1.5 | 1.5 | 2.5 | 3 |
| Network monitoring (12) | 1 | 1 | 2 | 2.5 |
| Application and service monitoring (10) | 1 | 2 | 3.5 | 4 |
| Data collection and scale (12) | 1 | 2 | 4.5 | 5 |
| Alerting and event management (12) | 2 | 2 | 3.5 | 4 |
| Visualisation and reporting (8) | 3.5 | 4 | 4.5 | 4.5 |
| Logs and traces (8) | 0.5 | 3.5 | 4.5 | 5 |
| Automation and remediation (8) | 2.5 | 3 | 3.5 | 4 |
| Security and compliance (8) | 2.5 | 3.5 | 3.5 | 4 |
| New-platform support (VMware exit) (10) | 1 | 1.5 | 2.5 | 3.5 |
| Total / 100 | 31.6 | 45.0 | 67.6 | 77.8 |
Year by year: 2016: 31.6 · 2017: 31.6 · 2018: 31.6 · 2019: 45.0 · 2020: 46.0 · 2021: 49.6 · 2022: 67.6 · 2023: 67.6 · 2024: 73.0 · 2025: 73.0 · 2026: 77.8
What moved the score
| Year | Domain | Change | Points | Trigger |
|---|---|---|---|---|
| 2019 | Application and service monitoring | 1 → 2 | +2.0 | 6.0: Explore correlates app metrics with Loki logs; probes via third-party blackbox exporter |
| 2019 | Automation and remediation | 2.5 → 3 | +0.8 | File provisioning of dashboards and data sources; official Terraform provider |
| 2019 | Data collection and scale | 1 → 2 | +2.4 | Loki 1.0 log store; metrics collection and storage still external |
| 2019 | Logs and traces | 0.5 → 3.5 | +4.8 | 6.0 with Loki 1.0: label-indexed log ingestion, LogQL search, live tail |
| 2019 | New-platform support (VMware exit) | 1 → 1.5 | +1.0 | Kubernetes through community kube-prometheus-stack; no hypervisor integrations |
| 2019 | Security and compliance | 2.5 → 3.5 | +1.6 | 6.x Enterprise: SAML, team sync, data source permissions, audit logging |
| 2019 | Visualisation and reporting | 3.5 → 4 | +0.8 | 6.x: new grid, Explore, gauges; Enterprise scheduled PDF reports |
| 2020 | Application and service monitoring | 2 → 2.5 | +1.0 | 7.0: Jaeger and Zipkin trace data sources |
| 2021 | Alerting and event management | 2 → 3.5 | +3.6 | 8.0: unified alerting, Alertmanager-compatible grouping, routing trees and silences across data sources |
| 2022 | Application and service monitoring | 2.5 → 3.5 | +2.0 | Tempo 2.0 tracing with TraceQL; metrics, logs and traces linked in Explore |
| 2022 | Automation and remediation | 3 → 3.5 | +0.8 | Terraform provider 1.x, Ansible collection, webhooks into AWX and runbook tools |
| 2022 | Data collection and scale | 2 → 4.5 | +6.0 | Mimir 2.0, 2022-03: HA multi-tenant TSDB on object storage, documented to 1 billion series |
| 2022 | Logs and traces | 3.5 → 4.5 | +1.6 | Tempo 2.0: trace IDs in Loki lines open traces; metrics, logs and traces correlated |
| 2022 | Network monitoring | 1 → 2 | +2.4 | Grafana Agent embeds snmp_exporter for interface counters; no traps, topology or NCM |
| 2022 | New-platform support (VMware exit) | 1.5 → 2.5 | +2.0 | Grafana Agent operator for Kubernetes; Proxmox and vSphere via third-party exporters; cloud data sources |
| 2022 | Server, virtualization and storage | 1.5 → 2.5 | +2.4 | Grafana Agent with embedded node and Windows exporters; hypervisors still need third-party exporters |
| 2022 | Visualisation and reporting | 4 → 4.5 | +0.8 | 9.x: Geomap, Canvas, node graph, heatmap, library panels; per-organization dashboards |
| 2024 | Application and service monitoring | 3.5 → 4 | +1.0 | 11.x: OpenTelemetry intake, service topology and RED views, Pyroscope profiling, Alloy blackbox probes |
| 2024 | Data collection and scale | 4.5 → 5 | +1.2 | Alloy 1.0 clustered collectors with Mimir: distributed collection, HA and long-term object storage |
| 2024 | Logs and traces | 4.5 → 5 | +0.8 | 11.0: OpenTelemetry logs and traces across the stack; Explore Logs app; correlations engine |
| 2024 | Network monitoring | 2 → 2.5 | +1.2 | Alloy 1.x: SNMP components and NetFlow, sFlow, IPFIX intake; Canvas maps drawn by hand; no NCM or IPAM |
| 2024 | Server, virtualization and storage | 2.5 → 3 | +1.2 | Alloy 1.0: first-party host, container and storage-array collection; curated infrastructure dashboards |
| 2026 | Alerting and event management | 3.5 → 4 | +1.2 | 12.4: SLO burn-rate alerting and federated alerting across instances; no topology root cause |
| 2026 | Automation and remediation | 3.5 → 4 | +0.8 | 13.x: grafana-operator and Crossplane GitOps; ServiceNow and Jira sync; remediation stays external |
| 2026 | New-platform support (VMware exit) | 2.5 → 3.5 | +2.0 | Alloy 1.x curated integrations for Proxmox VE, Nutanix AHV, KubeVirt and Kubernetes |
| 2026 | Security and compliance | 3.5 → 4 | +0.8 | 13.x Enterprise: RBAC, FIPS 140-2, Vault secrets, air-gapped licence (4.5) less 0.5 for CVE-2021-43798 on KEV 2025-10-09 (the security record) |
Direction, 2023 to 2026
Each item carries one theme and one driver (V vision, C customer need, M market window, U upstream, P portfolio).
| Date | Item | Theme | Driver |
|---|---|---|---|
| 2023-03 | Acquired Pyroscope; continuous profiling joins the stack | D3 Applications and services | P |
| 2023-06 | 10.0: correlations engine GA; subfolders | D8 Logs and traces | C |
| 2023-11 | 10.2: public dashboards GA | D6 Visualisation and reporting | C |
| 2024-04 | Alloy 1.0 OpenTelemetry Collector distribution replaces Grafana Agent | D4 Collection and scale | U |
| 2024-04 | Sift automated investigations | D7 AI-assisted operations | V |
| 2024-05 | 11.0: Scenes dashboards GA; Explore Metrics and Explore Logs | D6 Visualisation and reporting | C |
| 2024-08 | 11.2: query caching for self-hosted Enterprise | D4 Collection and scale | C |
| 2024-08 | USD 270M investment round led by Lightspeed | D12 Packaging and licensing | P |
| 2025-01 | mcp-grafana v0.1.0 preview | D7 AI-assisted operations | M |
| 2025-05 | 12.0: natural-language query assistant; AI cluster dashboards | D7 AI-assisted operations | M |
| 2026-02 | 12.4: SLO tracking, burn-rate and federated alerting | D5 Alerting and events | C |
| 2026-04 | 13.0: Scenes 2.0; MCP tools in core | D6 Visualisation and reporting | V |
| 2026-07 | mcp-grafana v1.0.0 GA | D7 AI-assisted operations | M |
| 2026-08 | 13.2: streaming dashboards; root-cause clustering | D6 Visualisation and reporting | C |
- Centre of gravity: Visualisation and reporting (4), AI-assisted operations (4), Collection and scale (2)
- Driver mix: Vision 14%, Customer need 43%, Market window 21%, Upstream 7%, Portfolio 14%
Credibility: 22.0 / 25
| Criterion | Score | Basis |
|---|---|---|
| Cadence | 5 | Monthly minor releases and a yearly major at GrafanaCON, kept since 2016 |
| Roadmap transparency | 4.5 | Public GitHub issues and code; direction at GrafanaCON and ObservabilityCON |
| Say-do | 4.5 | 9 of 9 items since 2023 shipped; public dashboards 5 months late, AI assistant 2 months late |
| Velocity | 5 | Several substantive capabilities per release across six products |
| Lifecycle stability | 3 | Legacy alerting and AngularJS removed; Grafana Agent replaced by Alloy; short support windows |
Say-do record, 2023 to 2026: 9 items 2023-2026, all shipped; 7 on time, public dashboards GA 5 months late (10.2), natural-language query assistant 2 months late (12.0).
Commercial risk: 9.5 / 25 (lower is better)
| Criterion | Score | Basis |
|---|---|---|
| Price and licence volatility | 1.5 | AGPLv3 relicensing in 2021 predates the window; since 2023 the Agent-to-Alloy repackaging only |
| Purchase constraints | 2.5 | Open source free; Enterprise subscription only, quote based, from about USD 29,000 a year |
| Channel and access | 1 | Open downloads, docs and trials; partner programme not described in the research |
| Owner stability | 2 | Independent but venture-funded at a USD 6B valuation; growth pressure, no acquisition |
| Cost of staying supported | 2.5 | Enterprise data source plugins stop when the licence lapses; frequent upgrades and forced Agent migration |
Security record of the product itself: CVE-2021-43798 (path traversal, fixed 8.3.1 2021-12-07) added to KEV 2025-10-09: minus 0.5. CVE-2021-39226 on KEV 2022-08-25, outside the window. Source: the security record.
Technical lock-in: 5.5 / 25 (lower is better)
| Criterion | Score | Basis |
|---|---|---|
| Formats | 2 | Dashboards in Grafana's own JSON schema; metrics in open Prometheus TSDB and Parquet formats |
| Export path | 1.5 | Full JSON export through the API; data readable with PromQL and LogQL APIs; Enterprise plugins stop on expiry |
| Stack coupling | 0.5 | 100+ data sources; OpenTelemetry and Prometheus standards; object storage of choice |
| Hardware / cloud coupling | 0.5 | Any hardware, offline; Enterprise needs air-gapped licence activation |
| Skills and tooling coupling | 1 | PromQL and Grafana skills are widespread |
Integrator fit: 15.0 / 20
| Criterion | Score | Basis |
|---|---|---|
| Partner programme openness | 2.5 | Partner programme not described in the research |
| Multi-tenancy and self-service | 3.5 | Organizations in Grafana; per-tenant isolation in Mimir and Loki via X-Scope-OrgID; no billing portal |
| API and infrastructure-as-code | 5 | Full HTTP API; official Terraform provider with 45M+ downloads, Ansible collection, Kubernetes operator |
| Skills and certification | 4 | Grafana Certified Associate and Certified Operator exams |
AI leverage: 13 / 20
| Strand | Today | Latest step |
|---|---|---|
| AIOps on monitoring data | 3 | 2024: Sift automated investigations; documented for Cloud, counted half; GPU dashboards since 9.0 |
| AI platform stack | 3 | 2025: 12.0: AI assistant GA in Enterprise; AI cluster dashboards for LLM latency and GPU memory |
| AI-assisted operations | 3 | 2025: 12.0: natural-language PromQL/LogQL assistant GA in Enterprise; recommends, does not act |
| Agent openness | 4 | 2026: mcp-grafana v1.0.0 GA, 2026-07; MCP tools also in Grafana 13.0 |
AI say-do: 3 tracked, 2 delivered GA (Sift, MCP 1.0), 1 delivered as preview (assistant in 11.0), 0 dropped. AI recommends and writes queries; it does not act.
Reading
Grafana Labs is turning a visualisation tool into a full self-hosted observability stack, with Mimir, Loki, Tempo and Alloy for collection and storage, and AI assistance and MCP on top. Its release rhythm and say-do record are the strongest of this group, though upgrades regularly remove older features. For classic infrastructure work the gaps are network management, hypervisor discovery and topology root cause, and several AI features are documented for Cloud first.
Open questions and evidence caveats
- Grafana ML, Sift and Application Observability are cited partly from Grafana Cloud docs; self-hosted Enterprise availability needs confirming.
- Scores assume the whole stack is deployed; Grafana alone collects nothing.
- Grafana OnCall and enterprise partner details are absent from the research.
Related reading
More from Infrastructure Monitoring Ledger 2026
- Overview: all platforms side by side
- Infrastructure Monitoring Ledger Methodology: How Every Score Is Built
- Broadcom DX NetOps and DX UIM: Monitoring Ledger 2026
- Centreon: Monitoring Ledger 2026
- Checkmk: Monitoring Ledger 2026
- Elastic Observability: Monitoring Ledger 2026
- IBM SevOne / Netcool: Infrastructure Monitoring Ledger 2026
- Icinga: Infrastructure Monitoring Ledger 2026
- LibreNMS: Monitoring Ledger 2026
- ManageEngine OpManager Nexus: Monitoring Ledger 2026
- Microsoft System Center Operations Manager: Monitoring Ledger 2026
- Nagios XI: Monitoring Ledger 2026
- Netdata Agent with self-hosted Parents: Monitoring Ledger 2026
- OpenNMS Horizon and Meridian: Monitoring Ledger 2026
- Opsview Monitor: Monitoring Ledger 2026
- Progress WhatsUp Gold: Monitoring Ledger 2026
- Prometheus: Monitoring Ledger 2026
- PRTG: Infrastructure Monitoring Ledger 2026
- ScienceLogic Skylar One: Monitoring Ledger 2026
- SolarWinds Observability Self-Hosted: Monitoring Ledger 2026
- Zabbix: Monitoring Ledger 2026