Datadog alternatives: where to go and how to switch without losing your alerts
Six tools in depth, how the Datadog bill really adds up, and a seven-step switch plan.
The short answer: Grafana Cloud if you want the biggest free tier and open standards (Prometheus, OpenTelemetry). New Relic if you want one platform billed per GB and per user instead of per host. SigNoz if you want an OpenTelemetry-native tool that ships a Datadog migration guide. Better Stack if you're a small team that also pays for an on-call tool. Dynatrace if you run hundreds of hosts and want auto-instrumentation. Elastic Observability if your team already lives in Elasticsearch. We compared these six tools in 2026 using each vendor's official pricing, their migration docs, and accounts from engineers who left Datadog.
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 end up leaving Datadog
Nobody leaves Datadog because it's a bad product. Even the angriest Hacker News threads give it credit for APM and for putting metrics, traces and logs in one UI. People leave over the bill, the sales process, and the fact that everything they build stays in Datadog's format. Six reasons keep coming up.
A bill with six meters
Infrastructure is billed per host ($15 a month on Pro with an annual commitment, $18 on-demand). APM is billed per host too ($31 annual, $36 on-demand). Logs cost $0.10 per GB ingested, then $1.70 per million indexed events ($2.55 on-demand). Custom metrics and synthetics come on top. Twenty hosts with infra and APM already run $920 a month before a single log line. List price never tells you the invoice.
Custom metrics that blow up
Datadog counts a custom metric as a metric name plus a unique combination of tag values, host included. Tag one metric with a 50-value customer_id across 20 hosts and you've got 1,000 series. Histograms and distributions count as 5 each. You get 100 per host on Pro (200 on Enterprise), pooled across the fleet, and the price for indexed metrics over that allotment isn't published: it's set in your contract.
Paying twice for logs
Every GB is billed at ingestion, then every million events you keep searchable is billed again at indexing, with 15-day retention by default. Datadog's answer is Flex Logs ($0.60 per million on Starter, $0.05 per million for storage up to 15 months). That's a third product to learn, not an automatic price cut.
An annual commitment that locks you in
On-demand pricing runs about 20% above annual ($18 vs $15 per host on Pro). So teams commit for a year, then pay any overage at the higher rate. Teams also complain there's no real spending cap. One HN commenter pointed out that hosts are billed by the hour while spot instances bill by the second.
A pushy sales process
It's the top complaint after price: calls to personal phones, mandatory check-ins with an account manager who doesn't know your setup, and discounts that only come if you buy more products. One engineer on HN summed it up: the worst thing about Datadog is the sales process.
Everything stays in Datadog's format
The agent, the dd-trace libraries, DogStatsD, dashboards, monitors and the query language are all proprietary. Nothing copies over as-is. That's why most alternatives bet on OpenTelemetry: once your code is instrumented with it, changing vendors no longer means touching code.
You're leaving because… go to…
The right Datadog alternative depends on what's pushing you out. Usually one line fits. Each card jumps to the full write-up below.
Free up to 10k active series, Pro from $19 a month plus usage.
You want all-in-one without per-host billingNew Relic100 GB free each month, then $0.40/GB plus per-user pricing.
You're already on OpenTelemetry, or want to beSigNozCloud from $49 a month, open-source edition, Datadog migration guide.
You're a small team also paying for PagerDutyBetter StackLogs, uptime and on-call in one tool.
You run hundreds of hosts and want less manual instrumentationDynatraceFull-Stack at $58 a month per 8 GiB host, annual commitment.
Your team already uses Elasticsearch or KibanaElastic ObservabilityServerless, log ingest from $0.07/GB.
Datadog alternatives, tool by tool
For each tool: why move to it, who it's for, its limit, how to migrate from Datadog, and what it costs. The side panel always shows the same four markers: free plan, billing unit, starting price and main limit. Prices come from official pricing pages in 2026, in US dollars.
Grafana Cloud: the most generous free tier, built on open standards
Open standardsWhy switch to Grafana Cloud. Grafana Cloud runs on open-source building blocks: Prometheus and Mimir for metrics, Loki for logs, Tempo for traces, Pyroscope for profiling. Anything you build there can run on your own hardware later, which breaks the lock-in that pushes people off Datadog. The free tier is the biggest on this list: 10k active series, 50 GB of logs, 50 GB of traces, 50 GB of profiles, 14-day retention and 3 users.
Who it's for. Engineering teams comfortable with PromQL, often on Kubernetes, who want usage-based pricing and a self-hosted fallback.
Who it's not for. Teams that want APM as guided as Datadog's without learning a new query language. One engineer who moved from Datadog to Grafana put it bluntly on HN: don't do it unless they've got you over a barrel on pricing.
Its limit. The Pro bill has several lines: $19 a month base with 10k series included, then $6.50 per 1,000 series, logs and traces billed for processing ($0.05/GB), writes ($0.40/GB) and retention ($0.10/GB), plus $8 per active user. Enterprise needs a minimum commitment of $25,000 a year.
How to migrate. Grafana ships a Datadog dashboard translator. It converts screenboards into timeboards and remaps metric names (dots become "_dot_"), but it doesn't carry over alerts, and users on Grafana's forum report dashboards that didn't translate cleanly. Grafana's Datadog metrics proxy has been deprecated since 2024, so route through the OpenTelemetry Collector instead (step 3 of the plan).
- Free plan
- yes, 10k series, 50 GB logs
- Billed per
- series and GB
- Starting price
- $19/mo + usage
- Limit
- multi-line Pro bill
New Relic: all-in-one, billed per GB instead of per host
All-in-oneWhy switch to New Relic. New Relic covers the same ground as Datadog (infrastructure, APM, logs, browser, synthetics) but doesn't bill per host or per metric. You pay for data ingested and for users. The free tier includes 100 GB a month, one full user and unlimited basic users. Past that, it's $0.40/GB ($0.60 with Data Plus). That customer_id tag stops being a budget problem.
Who it's for. Teams of 1 to 5 engineers who live in the tool daily, with lots of hosts or containers but controlled data volume.
Who it's not for. A 15-developer team where everyone wants full access: per-user pricing takes over.
Its limit. Seats. A Core user costs $49 a month. A full platform user on Standard is $10 for the first, then $99, capped at 5; on Pro it's $349 a month billed annually ($418.80 monthly). Data stored in the EU data center costs an extra $0.05/GB.
How to migrate. New Relic takes OpenTelemetry in. Start with a single service: dual-ship its traces to Datadog and New Relic for two weeks, then compare the projected per-GB bill with your Datadog APM line. Hand out free basic seats widely and keep full seats for whoever runs incidents.
- Free plan
- yes, 100 GB/mo, 1 full user
- Billed per
- GB and user
- Starting price
- $0.40/GB past 100 GB
- Limit
- up to $349 per Pro user
SigNoz: the alternative built for OpenTelemetry
OpenTelemetry-nativeWhy switch to SigNoz. SigNoz puts metrics, traces and logs in one UI, built on OpenTelemetry from day one. There's a free community edition you host yourself and a cloud version with US, EU and India regions. The team publishes a full Datadog migration guide and a tool that translates your dashboard JSON.
Who it's for. Startups and mid-size engineering teams already instrumenting with OpenTelemetry, or using the switch as the moment to adopt it, who want a readable per-GB bill.
Who it's not for. Large enterprises that need Datadog's full breadth (security products, advanced synthetics, hundreds of turnkey integrations).
Its limit. The Teams plan is $49 a month with $49 of usage included (about 163 GB of logs or traces). After that: $0.30/GB for logs and traces (15 days), $0.10 per million metric samples (1 month). There's a startup rate of $19 a month for 12 months, and Enterprise starts at $4,000 a month. Self-hosting the open-source edition costs engineering time.
How to migrate. The SigNoz guide follows the safest order: metrics, then traces, then logs, using the Datadog receiver in the OpenTelemetry Collector to bridge without code changes. Pick the EU region at signup if your data has to stay in Europe.
- Free plan
- self-hosted open source
- Billed per
- GB and sample
- Starting price
- $49/mo, usage included
- Limit
- narrower scope than Datadog
Better Stack: logs, uptime and on-call for small teams
Logs + on-callWhy switch to Better Stack. Better Stack, founded in Prague in 2021, bundles uptime monitoring, status pages, on-call and logs. For a small team it often replaces two bills: Datadog and PagerDuty. The founder of a podcast search engine wrote on HN that he moved to Better Stack for exactly that, with no need for a constant stream of new features. The free plan gives you 10 monitors, one status page, 3 GB of logs and 3 GB of traces (3 days), 30 GB of metrics and 100,000 exceptions.
Who it's for. Teams of 2 to 10, SaaS products that need a simple status page and on-call rotation, and anyone who prefers a European vendor.
Who it's not for. Teams that depend on Datadog APM to chase latency across 40 microservices: Better Stack doesn't go as deep there.
Its limit. A responder license (the on-call person) costs $34 a month, $29 billed annually; other team members are free. In the Europe region, logs cost $0.10/GB and retention $0.05/GB per month. Europe bundles run from Nano ($30 a month, $25 annually, 40 GB of each signal) to Tera ($500, $420 annually).
How to migrate. Start with on-call and uptime, the easiest pieces to move. Rebuild your Datadog synthetic tests as monitors. Then dual-ship logs for one billing cycle. If uptime is all you need, also see our guides to switching from Pingdom and switching from UptimeRobot.
- Free plan
- yes, 10 monitors, 3 GB logs
- Billed per
- GB and responder
- Starting price
- $29/mo annual (on-call)
- Limit
- shallower APM
Dynatrace: auto-instrumentation for large fleets
EnterpriseWhy switch to Dynatrace. Dynatrace installs a single agent that discovers and instruments processes, services and their dependencies on its own. Across several hundred hosts, that's a lot of tracing libraries you no longer maintain. Its pricing is public and broken down by the hour, and Dynatrace says there's no overage penalty on its annual commitment.
Who it's for. Enterprises and IT departments that want one contract, many hosts, and little engineering time spent on instrumentation.
Who it's not for. A 10-host startup: the minimum annual commitment and contract model are built for big volumes.
Its limit. Full-Stack costs $58 a month for an 8 GiB host ($0.01 per GiB-hour, so a 32 GiB host costs roughly four times as much). Infrastructure-only is $29 per host per month, Foundation $7. Logs and traces: $0.20/GiB. You need a minimum annual commitment; a 15-day trial lets you check first.
How to migrate. Install the Dynatrace agent next to the Datadog agent on one host group, let auto-discovery run for a week, then compare the service map with your current APM. Get your commitment quoted on actual host memory, not host count.
- Free plan
- no, 15-day trial
- Billed per
- host memory, hourly
- Starting price
- $58/mo per 8 GiB host
- Limit
- minimum annual commitment
Elastic Observability: the natural next step if you run Elasticsearch
ELK teamsWhy switch to Elastic. If your team already searches in Kibana, Elastic Observability adds metrics, APM and traces on the same engine. The serverless version bills per GB, with no hosts or custom metrics to count, and log search remains its strongest suit.
Who it's for. Teams with an ELK history and heavy log volume who want fast full-text search over long histories.
Who it's not for. A small team with no Elasticsearch experience thinking of self-hosting: running it well takes real operational skill.
Its limit. On serverless, Logs Essentials bills ingest from $0.07/GB and retention at $0.017/GB per month. Complete bills log and trace ingest from $0.09/GB, metrics at $0.023/GB, and retention at $0.019/GB per month, with 50 GB of free egress. Elastic has announced new metrics pricing for 2026, so check it when you sign.
How to migrate. Logs first: dual-ship from the OpenTelemetry Collector or Elastic Agent and compare your ingest cost with Datadog's two log lines (ingest plus indexing). Metrics and traces follow.
- Free plan
- no
- Billed per
- GB ingested and retained
- Starting price
- $0.07/GB (logs)
- Limit
- demanding to self-host
The 6 alternatives vs Datadog
For a reference point, take 20 hosts with infrastructure and APM. On Datadog, billed annually, that's $300 for infra and $620 for APM: $920 a month before logs and custom metrics. On New Relic, host count doesn't matter: you pay for GB past 100 and for your users. On Dynatrace, 20 Full-Stack hosts at 8 GiB come to $1,160 a month. One row per tool: what it replaces on the left, starting price and migration path on the right.
| What it replaces | Starting price and migration | |
|---|---|---|
| DatadogWhere you start | Infrastructure, APM, logs, synthetics, security | $15/host (infra Pro, annual) + $31/host (APM) + $0.10/GB of logs, then indexing |
| Grafana CloudOpen standards | Metrics, logs, traces, profiles | Free up to 10k series, Pro $19/mo + usage; dashboard translator |
| New RelicAll-in-one | Infrastructure, APM, logs, synthetics | 100 GB free, then $0.40/GB + users; dual-ship via OpenTelemetry |
| SigNozOpenTelemetry-native | Metrics, traces, logs | $49/mo with usage included; Datadog migration guide and tool |
| Better StackSmall teams | Logs, uptime, on-call, status pages | Free, on-call $29/mo annual; move on-call first |
| DynatraceLarge fleets | Infrastructure, APM, logs, auto-discovery | $58/mo per 8 GiB host; agent installed next to Datadog |
| Elastic ObservabilityELK teams | Logs, metrics, APM | From $0.07/GB ingested; logs first |
Prices exclude tax, in US dollars, as listed by each vendor in 2026 and subject to change. Every vendor bills a different unit, so run the numbers on your own volume.
Switching from Datadog in seven steps
Order matters. Each step protects something: your history, your alerts, your code and your contract. Durations are rough estimates. Plan on 30 to 90 days of running both for a mid-size team, much longer for a large one.
Take inventory and set the end date
Month 1Export every dashboard as JSON from the Datadog UI, and list your monitors, integrations and synthetic tests. Write down your annual contract's renewal date: it sets your timeline. Starting three months out gives you room to run both without paying for another year.
Tip: sort dashboards by last viewed. Anything nobody has opened in six months doesn't migrate.
- Time
- 2 to 3 days
- Protects
- your contract and history
Point the new tool at one service
Month 1Pick a representative service with real traffic. Open a free account (Grafana Cloud, New Relic, Better Stack) or the destination's trial and run it alongside Datadog. Check three things: does latency show up the same, are logs readable, does the projected bill hold?
- Time
- 1 to 2 weeks
- Cost
- often $0
Bridge with OpenTelemetry, no code changes
Months 1-2The OpenTelemetry Collector has a Datadog receiver: it accepts traffic from the Datadog agent and dd-trace libraries and forwards it to the new tool. Configure it to dual-ship to Datadog and the destination. You keep a rollback path at all times.
Watch out: Datadog's own docs say a misconfigured OpenTelemetry receiver can get its metrics classified as custom, with extra charges. Keep an eye on your usage page while you dual-ship.
- Time
- a few days
- Protects
- your code, untouched
Move one signal at a time
Month 2Metrics first, then traces, then logs. Replace the Datadog agent with the OpenTelemetry Collector, DogStatsD with the OpenTelemetry SDK, and dd-trace with each language's OpenTelemetry SDK. Rename tags: service becomes service.name, env becomes deployment.environment. Go service by service, never the whole fleet at once.
- Time
- 2 to 6 weeks
- Order
- metrics, traces, logs
Translate dashboards, then fix them by hand
Month 2Run Grafana's translator or SigNoz's tool on your JSON exports. Expect cleanup: metric names change and some widgets have no equivalent. Start with your 10 most-viewed dashboards, on-call ones first.
- Time
- 1 to 3 weeks
- Automated
- only partly
Rebuild monitors one by one and check parity
Months 2-3No tool migrates Datadog monitors automatically. Rebuild them by hand, then run both sides in parallel. For each alert, ask: did it fire at the same moment on both? A different query language or shorter retention creates silent gaps, and you only find them at the first missed incident.
- Time
- 2 to 4 weeks
- Risk if rushed
- an incident nobody hears about
Cut Datadog off before renewal
Contract endOnce every alert has proven itself, stop dual-shipping, uninstall the Datadog agent and remove the dd-trace libraries. Give Datadog written notice before the renewal date in your contract. Keep your JSON exports as a record in case an old incident comes back up.
- Time
- 1 week
- Protects
- your next renewal
Is your tool missing from this page? Email us: hello@hackceleration.com
What makes a Datadog exit go wrong
The same mistakes show up again and again in threads from teams who left Datadog. None is hard to avoid, but each one bites at the worst time: mid-incident.
Cutting over before alerts are proven
The most expensive mistake. A rebuilt alert with the wrong threshold or window stays silent, and nobody notices until the outage. Keep both running until every alert has fired at least once on each side.
Comparing list prices
Datadog's per-host price against someone else's per-GB price tells you nothing. Take your last Datadog invoice line by line and price the same volume at the destination: GB of logs, series count, user count.
Recreating the cardinality that cost you
If a customer_id or user_id tag blew up your custom metrics on Datadog, it'll blow up your active series on Grafana Cloud too. Use the switch to drop it or move it into traces.
Moving everything at once
Swapping agent, libraries, dashboards and alerts in the same week makes it impossible to tell where a gap came from. One service, then one signal, then the next.
Underestimating self-hosting
An open-source edition has no license fee, but it costs engineers to run, upgrade and back up. Price that time before you self-host to save money.
Missing the renewal date
An annual contract that auto-renews for lack of notice means another year of Datadog on top of the new tool. Log the notice date during inventory.
- Dashboards exported as JSON, monitors and integrations listed
- Datadog renewal and notice dates written down
- New tool trialed on one real service for one to two weeks
- OpenTelemetry Collector dual-shipping, Datadog usage page watched
- Tags renamed (service.name, deployment.environment)
- Top 10 dashboards translated and cleaned up
- Every monitor rebuilt and fired at least once on both sides
- Datadog agent and dd-trace libraries removed
- Written notice sent before renewal
Datadog alternatives: frequently asked questions
What is the cheapest Datadog alternative?
For a small team, Grafana Cloud's free tier (10k active series, 50 GB of logs, 50 GB of traces) or New Relic's (100 GB a month, one full user) can cost $0. For heavy log volume, Elastic serverless starts at $0.07/GB ingested. The cheapest option is the one whose billing unit matches your shape: lots of hosts and few users favors New Relic, lots of logs favors Elastic or Better Stack.
Is there a free, open-source alternative to Datadog?
Yes. SigNoz has a free community edition that puts metrics, traces and logs in one UI on OpenTelemetry. The Grafana stack (Prometheus or Mimir, Loki, Tempo) is the modular route. Coroot, which uses eBPF to observe services without code changes, has a free community edition too. The license is free; running it is not.
Why is Datadog so expensive?
Because each signal has its own meter: hosts for infrastructure, hosts again for APM, GB and then million events for logs, and tag combinations for custom metrics. A histogram counts as 5 metrics. Tip: sort your usage page by metric to find the few that drive the overage, then strip their high-cardinality tags.
Datadog vs New Relic: which should you pick?
New Relic if you have many hosts and few users: it doesn't bill per host or per metric, only GB past 100 a month and full seats. Datadog stays more comfortable if a large team uses the tool every day, since New Relic's full Pro seat runs $349 a month.
Datadog vs Grafana: which is better?
Grafana if you want open standards, a big free tier and the option to self-host. Datadog if your team wants guided APM without learning PromQL and can live with the bill. A middle path: keep Datadog for APM and move metrics and logs to Grafana, which cuts the bill without redoing everything.
Can I keep using OpenTelemetry with Datadog?
Yes. Datadog accepts OpenTelemetry three ways: its agent with the embedded collector, an OpenTelemetry Collector exporting OTLP, or direct OTLP ingest. Its docs say metrics from supported receivers carry no extra cost, but a misconfigured receiver gets them counted as custom metrics. Instrumenting with OpenTelemetry is mostly what makes your next migration easy.
How long does a Datadog migration take?
Thirty to 90 days of running both for a mid-size team, and 6 to 12 months for a large organization. The slow part isn't instrumentation, it's rebuilding and verifying monitors one by one. Start three months before your renewal date.
What about error tracking and tracing-only tools?
If you mostly used Datadog for errors, Sentry's Team plan costs $29 a month ($26 annually), with a free Developer plan. If deep distributed tracing is the main need, Honeycomb's free plan covers 20 million events a month and Pro starts at $150. Both are specialist tools: pair them with one of the six above for infrastructure. For incident management on the IT side, see our ranking of ITSM software.
When should you stay on Datadog?
When your bill is stable, your team really uses APM, security and synthetics in one place, and nobody has three months to run a migration. Negotiate instead: move rarely-read logs to Flex Logs, drop high-cardinality tags, and tune trace sampling. Datadog also has an EU1 region hosted in Germany if data location is the only issue. Moving cloud providers too? See our guides to switching from AWS and switching from Vercel.
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.