Grafana alternatives: where to go, and how to switch without losing your alerts
Eight tools in depth, the Grafana Cloud bill explained, and a six-step switch plan.
The short answer: pick SigNoz if you want one open-source tool for metrics, logs and traces instead of the whole LGTM stack. Pick Datadog if you never want to run observability infrastructure again. Pick New Relic if your small team fits in 100 GB free per month and wants to keep its PromQL queries. Pick Elastic if searching logs is what hurts most. If only the metrics backend is the problem, keep Grafana on screen and swap what's behind it (VictoriaMetrics). Want zero-config server monitoring? Netdata. And if your dashboards track revenue rather than CPU, you need a BI tool: Metabase or Superset. We compared all eight and checked their official pricing pages in 2026.
Some links may be affiliate links. The ranking is editorial: a sponsored spot is always labeled on its card, and no affiliate link changes a score.
Why teams leave Grafana, and what actually bothers them
Credit where it's due: Grafana is free to self-host, the Grafana Cloud free tier is never billed, and plenty of teams move to Grafana from Datadog to cut costs. People rarely leave because of the charts. They leave because of everything around them: the Cloud bill, the stack you have to run, log search, and upgrades.
A Grafana Cloud bill that's hard to predict
Pro starts at $19 a month plus usage, and usage is counted several ways. Metrics are billed per active series ($6.50 per 1,000 beyond the 10,000 included), measured at the 95th percentile of the month. Logs, traces and profiles are billed three times: per GB processed ($0.05), per GB written ($0.40) and per GB retained ($0.10). The classic trap is a high-cardinality label like user_id, which multiplies series. The quieter trap: a series scraped every 30 seconds costs twice as much as one scraped every 60, because Grafana also counts data points per minute.
The full LGTM stack is heavy to self-host
Self-hosting Grafana alone is easy. Self-hosting Mimir, Loki and Tempo is not. Mimir runs as microservices: distributor, ingesters, queriers, store-gateway, compactor. User reviews tell the same story: deployment goes fine, day-to-day management is the hard part. A small team can end up maintaining monitoring infrastructure nearly as big as the system it watches.
Loki slows down on full-text search
Loki only indexes labels, not log content. That keeps it cheap, and it's also why searching for a specific string across a long time range drags. Users describe splitting queries into small time windows just to avoid timeouts. LogQL also gets called finicky, with vague error messages. If your team lives in its logs, this is often the real reason to move.
A steep learning curve
It's the most common complaint in user reviews: hard to learn, hard to set up, and the recurring question of how to get data in at all. Grafana collects nothing on its own. It displays what Prometheus, an agent or a database sends it. For a team new to monitoring, that's three tools to learn instead of one.
Dashboards that break on upgrades
The dashboard schema changes between versions. Amadeus, which runs more than 5,000 dashboards, saw upgrades break dashboards through schema changes, and that's how the Perses project started. Keeping dashboards in Git is possible, but many users find the GitOps setup needlessly complicated.
A forced alerting migration
Grafana 11 removed legacy alerting. You could only migrate to the new system up to version 10.4, so jumping straight to 11 without migrating means losing your alerts, and some servers refuse to start with the old setting still on. Add the fact that many users find alert creation fiddly, and it's a natural moment to ask whether you rebuild your alerts here or somewhere else.
You're leaving because… go to…
The right Grafana alternative depends on what's pushing you out. Usually only one row applies. Each card links to the full profile.
SigNoz puts all three signals in one open-source tool. Datadog runs everything for you, for a bigger bill.
Your Grafana Cloud bill keeps surprising youNew RelicYou pay per GB and per user, with 100 GB free a month, and your PromQL queries keep working.
Log search is too slowElasticFull-text search is its core strength, and its migration tool converts Grafana dashboards and alerts.
Only the metrics backend hurtsVictoriaMetricsKeep Grafana, your dashboards and your queries. Swap only the metrics storage.
You just want to see your servers nowNetdataInstall the agent and it finds the metrics itself. No exporters, no dashboards to build.
Your dashboards are really business reportsMetabase or SupersetRevenue, signups, orders: those are SQL questions, not time series.
Grafana alternatives, tool by tool
For each tool: why switch to it, who it's for, its main limit, how to move over from Grafana, and what it costs. All prices are in US dollars, exactly as each vendor lists them.
SigNoz: one tool to replace the whole stack
Open source, all-in-oneWhy switch to SigNoz. Grafana needs four pieces (UI, metrics, logs, traces). SigNoz ships them as one product, built natively on OpenTelemetry. There's no per-host pricing: you pay for what you send.
Who it's for. Engineering teams that want off the LGTM stack without moving to a proprietary platform, and that already use or plan to use OpenTelemetry.
Who it's not for. Teams that depend on dozens of Grafana plugins. The SigNoz ecosystem is smaller.
Its limit. The full self-hosted stack is resource-hungry. And there's no direct import of Grafana dashboard JSON, because the schemas differ.
How to switch. SigNoz publishes an official "Migrate from LGTM" guide. The good part: PromQL runs as-is in SigNoz panels, so you copy queries instead of rewriting them. Dashboard templates are importable. A community tool (unofficial) converts a Grafana JSON file and checks each query against your instance, which saves time if you review the output.
Pricing. The Community edition is free to self-host. Teams starts at $49 a month, including $49 of usage, then $0.30 per GB of logs or traces (15-day retention) and $0.10 per million metric samples. Tip: startups under 3 years old, under 30 people and under $6M raised pay $19 a month for 12 months. Enterprise starts at $4,000 a month.
- Free
- Community edition
- Paid
- Teams from $49/mo
- Best for
- leaving the full stack
- Limit
- dashboards rebuilt
Datadog: hand everything off, at a premium
Most completeWhy switch to Datadog. You stop operating anything. Infrastructure, APM, logs and 1,000+ integrations, all hosted and connected in one UI. For a team spending more time maintaining monitoring than using it, that's a real relief.
Who it's for. Teams with budget that would rather pay a vendor than keep an engineer part-time on Mimir and Loki.
Who it's not for. Teams leaving Grafana to cut costs. The opposite move, from Datadog to Grafana for price, is one of the most common in this space.
Its limit. The bill has many meters: per host, per product, per log volume. Custom metrics beyond the included quota are billed on usage with no public per-unit price. Complaints about cost and pushy sales come up again and again.
How to switch. There's no official Grafana-to-Datadog dashboard import. A community browser tool, still a work in progress, translates PromQL into Datadog queries. Datadog accepts OpenTelemetry (through its DDOT agent or an OTLP collector), so your instrumentation can stay vendor-neutral. Tip: the Datadog plugin for Grafana lets you view both sources side by side during the cutover.
Pricing. Infrastructure is free up to 5 hosts, with only 1 day of metric retention. Infrastructure Pro costs $15 per host per month billed annually ($18 on demand), with 100 custom metrics per host. Enterprise is $23 ($27 on demand). APM alone is $36 per host per month, annually. Logs start at $0.10 per GB ingested.
- Free
- up to 5 hosts
- Paid
- Infra Pro $15/host/mo
- Best for
- zero ops
- Limit
- many billing meters
New Relic: pay per GB and keep your PromQL
Most generous free tierWhy switch to New Relic. The model is easier to read than Grafana Cloud's: data per GB, full users per seat. No active series or DPM to babysit.
Who it's for. Small engineering teams that fit within 100 GB free a month and want to keep PromQL while they learn NRQL.
Who it's not for. Large teams where everyone needs full access. Users are the expensive part.
Its limit. A Pro full-platform user costs $349 a month, billed annually. Standard caps you at five full users. Basic users are free but limited.
How to switch. It's the gentlest exit on this list. Prometheus 2.15+ ships metrics to New Relic via remote write, with no agent changes. The chart builder accepts PromQL-style queries and translates them to NRQL. Better still, New Relic works as a Prometheus data source inside Grafana 6.7+, so your existing dashboards keep running while you rebuild at your own pace.
Pricing. 100 GB free per month, then $0.40 per GB ($0.60 with Data Plus), plus $0.05 per GB in the EU data center. Core users cost $49 a month. Full platform: Standard is $10 for the first user then $99, Pro is $349 annually ($418.80 monthly).
- Free
- 100 GB/month
- Paid
- $0.40/GB after that
- Best for
- small teams, soft landing
- Limit
- pricey full users
Elastic Observability: when logs come first
Best for log searchWhy switch to Elastic. Elastic does what Loki struggles with: finding any string in a mountain of logs, fast. Kibana covers metrics and traces too.
Who it's for. Teams already comfortable with ELK, or whose daily work is digging through logs during incidents.
Who it's not for. Small teams hoping to self-host without dedicating real hardware. The stack is resource-heavy.
Its limit. Self-hosted weight. Serverless removes that problem but bills on ingest and retention, which takes the same volume discipline Grafana Cloud does.
How to switch. Elastic is the only vendor here with an official Grafana converter. Its Observability Migration Platform CLI reads your Grafana dashboards and alerts (from files or the API) and recreates them in Kibana. A flag keeps queries in native PromQL. Migrated alerts arrive disabled, and panels it can't convert cleanly get flagged for review, so you know exactly what to redo by hand.
Pricing. On serverless, Logs Essentials starts at $0.07 per GB ingested and $0.017 per GB retained per month. The Complete tier starts at $0.09 per GB for logs and traces, and $0.023 per GB of metrics ingested. The first 50 GB of egress is free, then $0.05 per GB.
- Free
- 50 GB egress
- Paid
- from $0.07/GB ingested
- Best for
- log search
- Limit
- heavy self-hosted
VictoriaMetrics: keep Grafana, swap the engine
No new UI to learnWhy switch to VictoriaMetrics. Because you may not need to leave Grafana at all. If the pain is the metrics backend (running Mimir, or paying for series), VictoriaMetrics replaces it and speaks PromQL. Dashboards and queries stay put.
Who it's for. Teams attached to their Grafana dashboards, running a high volume of active series.
Who it's not for. Small teams. The entry Cloud tier already assumes real volume.
Its limit. You keep Grafana, so you keep its downsides: learning curve, alerting migrations, dashboard upkeep. You fix metrics cost and ops, nothing else.
How to switch. Point your metrics at VictoriaMetrics, then add it as a Grafana data source. Since it understands PromQL (and its MetricsQL extension), most panels render untouched. For logs, there's VictoriaLogs.
Pricing. VictoriaMetrics Cloud starts at $200 a month for 500,000 active series, with $200 of credit for 30 days to try it. VictoriaLogs also starts at $200 a month.
- Trial
- $200 credit, 30 days
- Paid
- from $200/mo
- Best for
- keeping Grafana dashboards
- Limit
- keeps Grafana's flaws
Netdata: monitoring with zero configuration
Fastest to set upWhy switch to Netdata. Install the agent and it auto-detects the machine's metrics and shows them in real time. No exporters to wire up, no dashboards to build from scratch.
Who it's for. Small teams running Linux servers, and homelabs that want to see what's happening right now.
Who it's not for. Teams that need long history and month-over-month analysis.
Its limit. Netdata is built for real-time, per-node visibility. For long retention, tools like Zabbix are a better fit.
How to switch. There's no dashboard import, but you barely need one: the agent builds its own views. The real work is recreating your alerts.
Pricing. Community is free for up to 5 nodes. Homelab is $90 a year for personal use. Business starts at $7 per node per month billed annually for infrastructure ($8.75 monthly), $15 with logs and traces ($18.75 monthly).
- Free
- 5 nodes
- Paid
- Business from $7/node/mo
- Best for
- Linux servers, real time
- Limit
- long retention
Metabase: when your dashboards are about customers, not CPUs
Best for business dashboardsWhy switch to Metabase. Grafana is built for time series. Answering "revenue by plan this quarter" in it means fighting the tool. Metabase connects to your SQL database and lets non-technical people ask those questions themselves.
Who it's for. Teams that bent Grafana into tracking sales, signups or orders.
Who it's not for. Infrastructure monitoring. It isn't an observability tool.
Its limit. Everything runs on SQL. No real-time metrics, no traces, no on-call alerting.
How to switch. Rebuild your business dashboards as questions on your database. It usually goes faster than expected, because those dashboards were awkward in Grafana to begin with. For a wider view, see our ranking of BI and data visualization tools.
Pricing. The open-source edition is free to self-host. Starter is $100 a month ($90 billed annually), 5 users included then $6 per user. Pro is $575 a month ($517.50 annually), 10 users included then $12 per user. Enterprise starts at $20,000 a year.
- Free
- open-source edition
- Paid
- Starter $100/mo
- Best for
- SQL business dashboards
- Limit
- no monitoring
Apache Superset and Preset: heavier SQL analytics
Best for SQL-heavy teamsWhy switch to Superset. If Metabase feels too simple for your analysts, Superset gives SQL-fluent teams more room: a full SQL editor and a wide range of charts. Preset is the managed version, so you skip running it.
Who it's for. Data teams that write SQL every day and want an Apache 2.0 tool, or a managed one through Preset.
Who it's not for. Non-technical teams that need self-serve reporting with no SQL.
Its limit. Self-hosting in production means maintaining Docker, Redis, Celery and Python. Preset removes that work, for a per-user fee.
How to switch. Same logic as Metabase: rebuild business dashboards as SQL datasets and charts, and leave infrastructure monitoring to an observability tool.
Pricing. Superset is free to self-host (Apache 2.0). Preset Starter is $0 for up to 5 users. Professional is $20 per user per month billed annually, or $25 monthly. Embedded analytics starts at $500 a month for 50 viewers.
- Free
- Superset, or Preset up to 5 users
- Paid
- Preset $20/user/mo
- Best for
- SQL-fluent data teams
- Limit
- ops-heavy self-hosted
All 8 alternatives side by side
Our default picks: SigNoz to replace the whole stack, New Relic for the softest landing. To see the other side of these moves, read our guides to switching from Datadog and switching from New Relic.
| Free tier | Entry paid plan and switching from Grafana | |
|---|---|---|
| Grafana CloudStarting point | 10,000 series, 50 GB logs, 14 days, 3 users | Pro from $19/mo plus usage |
| SigNozOpen-source all-in-one | Self-hosted Community edition | Teams from $49/mo; PromQL works, dashboards rebuilt |
| DatadogFully managed | Up to 5 hosts | Infra Pro $15/host/mo annually; no official import |
| New RelicPer GB | 100 GB/month | $0.40/GB; Prometheus source in Grafana during cutover |
| ElasticLogs | 50 GB egress | From $0.07/GB ingested; official dashboard and alert converter |
| VictoriaMetricsKeep Grafana | $200 credit, 30 days | From $200/mo; dashboards kept |
| NetdataReal time | 5 nodes | Business from $7/node/mo; auto-built views |
| MetabaseBusiness dashboards | Open-source edition | Starter $100/mo; rebuild in SQL |
| Superset (Preset)SQL analytics | Superset self-hosted, Preset up to 5 users | Professional $20/user/mo annually |
Entry prices in US dollars before tax, as listed by each vendor, subject to change. Grafana itself says its pricing model is changing, so recheck its page before you compare.
Switching from Grafana in six steps
Order matters. Each step protects something: your visibility into production, your alerts, your bill.
Inventory everything that runs
Week 1Export every dashboard as JSON, and list your alerts, data sources and contact points. Then look at active series: in Grafana Cloud, Cost Management shows historical cardinality, meaning which metrics create the most series. That's your list of what not to migrate as-is.
Tip: note which dashboards people actually open. The ones nobody visits don't need to move.
- Time
- one to three days
- Keep
- JSON exports
Make your collection vendor-neutral
Weeks 1 to 2Before switching tools, switch how you send data. The OpenTelemetry Collector works with SigNoz, Datadog, Elastic and New Relic, so changing destination becomes a config edit. If you stay on Prometheus, remote write does the same job for metrics.
- Standard
- OpenTelemetry
- Payoff
- new destination, no re-instrumenting
Dual-ship during the transition
Several weeksSend the same data to Grafana and to the new platform for several weeks. Compare numbers, confirm nothing is missing, and keep a trusted view if the new tool surprises you. Use this window to drop useless metrics: don't migrate your cardinality problem.
- Rule
- never cut the old one before comparing
- Bonus
- series cleanup
Convert or rebuild dashboards
Depends on volumeUse a converter when one exists: Elastic's official tool for Kibana, the Perses migration command, community converters for SigNoz or Datadog. Otherwise rebuild by hand, starting with your ten most-viewed dashboards. On SigNoz and New Relic, PromQL support saves a lot of time.
- Priority
- top ten dashboards
- Tools
- Elastic, Perses, community converters
Recreate and test every alert
Before cutoverThis is the step teams rush and the one that costs the most. Recreate each alert, then fire it on purpose to confirm it reaches the right place. On Elastic, imported alerts arrive disabled, so you turn them on one by one after checking them.
- Test
- fire each alert once
- Trap
- imported alerts are disabled
Cut over, then cancel
End of projectOnce alerts are validated and key dashboards are live, stop sending to Grafana. Keep the JSON exports somewhere safe. Then cancel Grafana Cloud Pro, or drop back to the free tier if you want access to whatever history it still retains.
- Keep
- exports and screenshots
- Option
- drop to Free instead of closing
Is your tool missing from this page? Email us: hello@hackceleration.com
What makes a Grafana switch fail
The same mistakes show up project after project. None of them is really technical. They're planning gaps.
Forgetting the alerts
Dashboards are visible. Alerts aren't. A team that moves its screens and shuts the old system down can lose a critical alert without noticing, until the day it should have fired. Write the full list before you start.
Migrating every metric
Copying your cardinality one-for-one just moves your bill to a new vendor. Use the switch to drop high-churn labels (user IDs, request IDs) and metrics nobody looks at.
Underestimating the query language change
Moving from LogQL or PromQL to another language takes time. Where possible, pick a destination that accepts PromQL (SigNoz, New Relic, VictoriaMetrics, Elastic with its native option), and budget training time for the rest.
Skipping the dual-ship period
Without a side-by-side comparison, you can't know the new platform sees everything. Run both until the numbers match over a normal week and through at least one incident.
Using an observability tool for business reporting
If half your dashboards are about sales or customers, don't move them to Datadog or SigNoz. Split the two jobs: monitoring in one tool, business dashboards in Metabase or Superset.
Overlooking the license if you modified Grafana
If you modified Grafana and ship it inside a product for customers, the AGPL weighs on your choice. For unmodified internal use, it changes nothing. Check which case you're in before you treat it as a reason to leave.
- Dashboards exported as JSON and ranked by views
- Alerts, data sources and contact points listed
- Active series and high-cardinality labels identified
- Collection moved to OpenTelemetry or remote write
- Dual shipping live and numbers compared
- Top ten dashboards rebuilt or converted
- Every alert recreated and fired once
- Grafana Cloud Pro canceled or downgraded, exports saved
Grafana alternatives: frequently asked questions
What is the best free, open-source alternative to Grafana?
It depends on the job. For full observability, SigNoz Community covers metrics, logs and traces in one free self-hosted tool. For quick server monitoring, Netdata is free up to 5 nodes. For business dashboards, Metabase and Superset both have free editions. And remember Grafana itself is free to self-host: often the cost lives in the rest of the stack, not the UI.
Is Perses a replacement for Grafana?
Not yet for most teams. Perses is a CNCF sandbox project since 2024, free, and built around dashboards as code, with a CLI command to migrate Grafana dashboards. It came out of Amadeus's struggle with Grafana dashboards breaking across upgrades. It's still young and has nothing equivalent to Grafana's Explore view, so treat it as one to watch if GitOps matters to you.
Does Grafana's AGPL license force me to open-source my code?
No, not for unmodified internal use. Grafana moved to AGPLv3 in 2021, and its CEO stated that unmodified distributions are not affected. The obligation applies if you modify Grafana and distribute it, or offer it as a network service. Little-known tip: Grafana publishes a free "Enterprise" binary under a proprietary license with the same features as the AGPL build.
How much does Grafana Cloud cost?
Pro starts at $19 a month plus usage: $6.50 per 1,000 metric series beyond 10,000, and for logs $0.05 per GB processed, $0.40 per GB written and $0.10 per GB retained. Each active user is $8 beyond the 3 included. Enterprise starts at a $25,000 annual commit. Grafana says its pricing model is changing, so check the current page. Tip: set up billing usage alerts on day one, and look at Adaptive Metrics, which Grafana says can cut metrics volume by up to 80%.
Grafana vs Kibana: which is better for logs?
Kibana, if you search free text in logs often. Elastic indexes log content, while Loki only indexes labels. That makes Loki cheaper but slower once a search goes beyond labels. Grafana keeps the edge for putting logs and metrics on one screen at lower cost.
Can I use Grafana for business dashboards?
You can, but you'll fight it. Grafana is built for values that change over time, like server load. For questions like "customers per plan this month," Metabase or Superset query your SQL database directly and stay readable for non-technical teams. If Tableau or Power BI are the tools you want out of, see our guides to switching from Tableau and switching from Power BI.
Can I keep Grafana and only change the backend?
Yes, and it's often the smartest money move. VictoriaMetrics replaces Mimir or Prometheus while speaking PromQL, and New Relic works as a Prometheus data source in Grafana. Your dashboards don't move. Grafana also plugs into automation tools: see how to connect it to n8n or to Claude through its MCP server.
Can I import Grafana dashboards into another tool?
Sometimes. Elastic officially converts Grafana dashboards and alerts to Kibana. Perses has a migration command. SigNoz needs a rebuild but runs your PromQL as-is. For Datadog there's no official path, only an unfinished community converter. Whatever you use, review every converted panel.
Get the next switch guide in your inbox
One tool to leave, one destination, one switch plan. Nothing else. Browse every guide in our switch tools section.