Datadog vs New Relic vs Grafana Cloud pricing 2027: host, GB or series

Summary. For a 50-node Kubernetes estate sending 2,000 GB of logs and 1,000 GB of traces a month, our worked bill at list prices checked on 13 November 2026 comes to about $3,550 a month on Grafana Cloud Pro, $4,850 on New Relic Pro with ten full users ($2,011 on Standard with five full and five core users) and $4,680 to $7,400 on Datadog's US site, depending on how many log events you index. Datadog's UK1 site lists 20% higher: $5,616 to $8,880. The drivers: Datadog's $15 per host for Infrastructure Pro, $31 for APM and $1.70 per million log events indexed; New Relic's $0.40 per GB after 100 GB free and $349 per full user on Pro; Grafana's $6.50 per 1,000 series and $0.45 per GB of logs or traces. Two dated changes matter: Grafana bills Kubernetes Monitoring and Application Observability plans started from 13 February 2026 on host hours with no telemetry credits, and Datadog replaced its free plan card with a contact-sales Basic tier in late October 2026. None of the three publishes sterling prices.

About this article. Written for engineering leads, SREs and FinOps teams in the US and UK preparing a 2027 observability renewal or migration. We checked every price, allotment and retention period against each vendor's pricing pages and billing documentation on 13 November 2026, and compared Datadog's and Grafana's pages with archived October 2026 copies. The worked bill uses our labelled assumptions and list prices, not negotiated rates. Vendors change prices often, so re-check before you commit; no vendor paid for inclusion. Researched with AI assistance and reviewed by our editorial team.

What to do before you sign a 2027 renewal

These three vendors do not agree on what they are selling. Datadog prices infrastructure and APM per host, then adds per-GB and per-event meters for logs and spans. New Relic charges per GB ingested and per engineer with full access, with no host charge. Grafana Cloud charges per active series, per GB at three pipeline stages, and per host hour for its Kubernetes and application products.

So the cheapest vendor depends on the shape of your telemetry. Before a renewal, export one month of five numbers: hourly node and container counts; active custom metric series; log volume in GB and in events, plus how far back engineers search; span volume in GB; and how many people troubleshoot versus only read dashboards. With those you can rebuild the worked bill below. If the renewal lands in the same quarter as a cluster upgrade, line the two up with our Kubernetes 1.35 end-of-life calendar, because node counts drive two of the three bills.

Two vendor statements need a label. Grafana's pricing page opens by declaring that charging more as customers send more data is broken and that Grafana is changing it. That is positioning: the page carries no date or rate comparison, and an archived copy from 3 October 2026 lists the same rates. New Relic's FAQ says you no longer count hosts or CPUs; you count gigabytes and engineers instead.

Two dated changes do appear in primary sources. Grafana bills Kubernetes Monitoring plans started after 13 February 2026 on active host and container hours with no telemetry credits; plans started between 17 September 2024 and that date keep credits of 1 billable series per host hour and 0.10 per container hour. Application Observability switched on the same date, and older plans pay a higher host-hour rate that includes 1.5 active series and 0.02 GB of traces per host hour. At Datadog, the pricing page captured on 15 October 2026 shows a free plan for up to five hosts with one-day metric retention; by 13 November that card is a Basic tier priced through sales, with 15-month retention and no container monitoring, though the FAQ still describes the free tier.

Pricing units at a glance

US list prices on 13 November 2026, with Grafana at its Pro starting rates.

Datadog: per host first, then per GB and per million events

Datadog's list page for its US site prices Infrastructure Pro at $15 per host per month billed annually and $18 month-to-month or on demand. Enterprise is $23 or $27, the DevSecOps tiers are $22 and $34 on annual terms, and Basic is priced through sales. A host is any physical or virtual operating system instance Datadog monitors, which on Kubernetes means the node, not the pod.

Each host licence carries allotments: Pro includes 5 containers, 100 custom metrics and 500 custom events per host; Enterprise includes 10, 200 and 1,000. Datadog counts these across the whole account rather than host by host. Extra containers cost $1 per container per month on annual or month-to-month terms, or $0.002 per container hour on demand, and extra custom metrics $5 per 100. Metric retention is 15 months.

APM lists at $31 per host per month on annual billing ($36 month-to-month) alongside infrastructure; APM Pro is $35 and APM Enterprise $40 on the same basis. Bought standalone, the three tiers are $36, $41 and $47. Each APM host includes 150 GB of ingested spans and 1 million indexed spans kept 15 days, averaged across APM hosts. Extra ingestion is $0.10 per GB; extra indexed spans cost $1.27, $1.70 or $2.50 per million for 7, 15 or 30 days.

Logs are where Datadog bills surprise people, because you pay twice. Ingest costs $0.10 per GB. Indexing then costs per million events, priced by retention: $1.06 for 3 days, $1.27 for 7, $1.70 for 15 and $2.50 for 30 on annual terms, rising to $3.75 on demand. Datadog suggests counting a day of events and multiplying by 30. Cheaper paths exist: Flex Logs storage at $0.05 per million events stored (query compute priced separately), Flex Logs Starter at $0.60 per million for 3 to 15 months, and archives in your own S3, Azure or Google Cloud bucket, with rehydration at $0.10 per compressed GB scanned plus indexing.

On-demand accounts get 15-day log retention by default. On a committed plan, volume above the commitment is billed at the annual price plus 50%. Volume and multi-year discounts start at 500 hosts a month, so a 50-node estate pays close to list.

The list page has a site selector. On 13 November 2026 every rate we use was 20% higher on UK1 than on the US site, still in dollars: Infrastructure Pro $18, APM $37.20, log ingest $0.12 per GB and 15-day indexing $2.04 per million events. EU1 matched US rates. AI is metered separately: AI credits list at $500 per 500-credit bundle on annual terms, and Agent Observability starts at $160 a month for 100,000 LLM spans, as our Datadog LLM Observability pricing guide explains.

Best for: teams that want one vendor for infrastructure, APM, security and incident tooling with unlimited user accounts, and the discipline to index only the logs they search.

New Relic: per GB plus per engineer

New Relic's default model has two meters. Data: the first 100 GB each month is free, then $0.40 per GB on the original data option or $0.60 on Data Plus, and storing data in the EU data centre adds $0.05 per GB. Users: basic users are free, core users cost $49 each, and full platform users, who get the whole platform, carry most of the bill.

On Standard, the first full platform user costs $10 and each additional user $99, with a cap of five. On Pro, a full platform user costs $349 a month with an annual commitment or $418.80 on monthly pay-as-you-go. Enterprise is quote-only. Hosts, agents and containers carry no extra charge.

A third model, New Relic Compute, replaces user licences with compute capacity units (CCUs) consumed by successful page loads and queries, so every engineer gets full access. Pro and Enterprise customers can choose it, but CCU rates go through sales, so we cannot price it.

Default retention on the original data option is short: 30 days for logs, 8 days for APM events and distributed traces, and 13 months for infrastructure integration data. Data Plus extends logs to 120 days and traces to 98, and 30 extra days across the organisation costs $0.05 per GB ingested. Data Plus also brings FedRAMP Moderate and HIPAA eligibility on Enterprise.

Watch the free tier ceiling: when an unpaid account passes 100 GB in a month, New Relic stops ingesting and blocks access until you upgrade or the month rolls over. It also says unused funds on Pro and Enterprise savings plans roll over for up to 12 months; confirm that in your order form.

Best for: teams with a small number of hands-on troubleshooters and many dashboard readers, or estates with many small hosts where per-host pricing would hurt.

Grafana Cloud: per series, per pipeline stage and per host hour

Grafana Cloud Free includes 10,000 active series, 50 GB each of logs, traces and profiles, three active users and 14 days of retention. Pro starts at $19 a month plus usage, with 13 months of metric retention and 30 days for logs and traces. Enterprise starts at a $25,000 a year commitment, with custom retention, Federal Cloud or Bring Your Own Cloud, and metrics as low as $3 per 1,000 series.

Pro rates are starting rates, with automatic volume tiers published behind each one:

• Metrics: $6.50 per 1,000 active series beyond the 10,000 included ($5.90 above 100,000).

• Logs and traces: $0.05 per GB processed, $0.40 per GB written ($0.37 above 1,000 GB) and $0.10 per GB retained, with 50 GB a month included.

• Kubernetes Monitoring: $0.01 per host hour ($7.20 a node a month) and $0.0007 per container hour ($0.50 a container), with 2,232 host hours and 37,944 container hours included.

• Application Observability: $0.025 per host hour ($18 a host a month), $0.023 above 25,000 host hours.

• Users: $8 per active user beyond three. Grafana Assistant: $20 per active AI user beyond three included, each with 40 million tokens. Agent monitoring is $1.50 per 1,000 generations, covered in our Grafana Cloud agent observability guide.

Metrics bill on active series at the 95th percentile, and Pro includes one data point per minute per series. A 15-second scrape, the Prometheus default, sends four, and billing then follows total data points.

Processed GB is everything that arrives, before Adaptive Telemetry drops anything; written GB is what is stored after optimisation, with a 50 GB allotment; retained GB is each 30-day increment beyond the 30-day minimum. With 30-day retention you pay process and write only, and queries are free up to 100 times your written volume. Traces use the same meters, and spans generated by Application Observability bill at trace rates.

For the host-hour products, Grafana's documentation says their metrics, logs, traces and profiles bill at standard rates on top of host hours. The pricing page, by contrast, describes Kubernetes Monitoring's included hours as covering the telemetry they generate. Check in the Cost Management and Billing app which series your Kubernetes integration bills. Hosts are metered only while active, so autoscaled nodes cost only the hours they exist.

Best for: teams already on Prometheus and OpenTelemetry who want pay-per-use billing, can manage cardinality, and value a free tier that a small service can live inside.

Head-to-head comparison

Rows draw on the pricing pages and on each vendor's site, retention and region documentation.

Worked monthly bill: a 50-node Kubernetes estate with logs and APM

The scenario is a mid-size US or UK SaaS company running 50 Kubernetes worker nodes all month and about 1,000 containers. APM or OpenTelemetry instrumentation covers services on all 50 nodes. Applications emit 20,000 custom metric series, scraped every 60 seconds. The estate ships 2,000 GB of logs a month, about 2 billion events at roughly 1 KB each, plus 1,000 GB of spans. Ten engineers troubleshoot and 40 people only read dashboards. We use a 30-day month (720 hours), list prices, no discounts and no tax, and assume indexed spans stay within Datadog's allotment.

Datadog US site, annual billing, Infrastructure Pro plus APM, all logs indexed for 15 days

Index only 20% of events (400 million, $680) and archive the rest, and the total falls to $4,680 plus storage. Month-to-month with every event indexed costs $8,480; 30-day indexing on annual terms costs $9,000. On UK1 every line is 20% higher: $8,880, or $5,616 with 20% indexed.

New Relic, original data option

The 500 GB for metrics and events is our assumption, because New Relic counts every signal in GB. On Standard, five full platform users ($10 + 4 × $99 = $406) plus five core users ($245) bring the total to $2,011, if the core users can work without full APM views. Pro pay-as-you-go is $5,548, Pro with Data Plus $5,530, and EU hosting adds about $170.

Grafana Cloud Pro, starting rates

We budget 30,000 series (the 20,000 application series plus 10,000 that Application Observability generates) and charge processing on every GB as a cautious reading. We subtract no Application Observability free hours, because the Pro card states none; with the free tier's 2,232 host hours, that line drops to $844.20. Grafana's published volume tiers would trim the total to roughly $3,380 to $3,480, under 5%, depending on how they are applied.

Side by side

Retention differs: Datadog's bill buys 15 days of logs and spans, New Relic's 30 days of logs but 8 of traces, Grafana's 30 days of both.

What moves the ranking:

• Log volume. With 1 KB events, fully indexed Datadog logs cost about $1.80 per GB for 15 days ($0.10 ingest plus $1.70 indexing), against $0.40 on New Relic and about $0.45 on Grafana. Larger events cut Datadog's per-GB figure; chatty one-line logs raise it.

• Cardinality. Datadog's custom metrics work out at $50 per 1,000 above the allotment, against Grafana's $6.50 per 1,000. A tag that multiplies series hurts far more on Datadog.

• Head count. Each extra troubleshooter costs $349 a month on New Relic Pro, $8 on Grafana and nothing on Datadog.

• Node count. Each extra node with APM adds $46 a month on Datadog's US site, $25.20 on Grafana and nothing on New Relic.

US and UK buying notes

Currency and tax. All three vendors publish list prices in US dollars only; on 13 November 2026 we found no sterling rate card on any of their pricing pages. Model a Datadog UK1 deployment on its UK1 rates, not the default US view. UK finance teams should agree invoicing currency, exchange-rate terms and VAT treatment in the order form.

Data residency. Datadog lists a UK1 site hosted in the UK and an EU1 site in Germany; sites are fully independent and cannot share data. Grafana Cloud offers UK stacks on AWS eu-west-2 and Google Cloud europe-west2 (the latter through Google Cloud Marketplace), but an existing stack cannot change region, so choose before you ingest. New Relic offers US and EU data centres, with EU storage adding $0.05 per GB, and names no UK region on its pricing page.

US public sector. Datadog describes its US1-FED site as FedRAMP High and says US2-FED is working towards IL5 authorisation. New Relic offers FedRAMP Moderate and HIPAA eligibility on Enterprise with Data Plus, and Grafana offers Federal Cloud on Enterprise. Check each vendor's trust documentation before sending regulated data.

Implementation checklist

1. Count events, not only GB. Datadog indexes per million events, so a 24-hour event count is the most useful number in a Datadog quote.

1. Set a cardinality budget per service. Series pricing ranges from $6.50 to $50 per 1,000 across these vendors, and on Grafana a 15-second scrape can quadruple the metrics line.

1. Decide retention per signal. Fifteen days of searchable logs covers many incident workflows; archive the rest to object storage you already pay for.

1. Instrument with OpenTelemetry. All three accept OpenTelemetry data. Datadog lists a managed OTLP intake at $0.50 per GB of ingested spans, so check which path you are billed on. For LLM spans, read our note on the OpenTelemetry GenAI conventions.

1. Dual-ship during evaluation. An OpenTelemetry Collector with two exporters sends the same telemetry to two backends for two to four weeks and gives you a real bill from each.

1. Check plan start dates. Grafana's host-hour rules depend on whether your plan started before or after 13 February 2026.

1. Negotiate at the right thresholds. Datadog volume discounts start at 500 hosts, New Relic offers commitment contracts on Pro and Enterprise, and Grafana Enterprise starts at $25,000 a year.

How eCorpIT can help

eCorpIT is a Gurugram-based technology organisation, founded in 2021, assessed at CMMI Level 5 and MSME certified, with senior-led engineering teams working across AWS, Microsoft and Google platforms. Our cloud computing services team can rebuild this bill from your own telemetry exports and set up OpenTelemetry collectors so you can dual-ship to two vendors before a 2027 renewal. Talk to us at /contact-us/.

Last updated: 13 November 2026.

References

1. Datadog Pricing List (US, EU1 and UK1 site rates, annual, month-to-month and on-demand) — Datadog

1. New Relic Pricing: editions, data and user costs, FAQ — New Relic

1. Grafana Cloud pricing: Free, Pro and Enterprise, product rates and volume tiers — Grafana Labs

1. Kubernetes Monitoring pricing — Grafana Labs documentation

1. Application Observability pricing — Grafana Labs documentation

1. Datadog Pricing, archived capture of 15 October 2026 — Internet Archive

1. Grafana Cloud pricing, archived capture of 3 October 2026 — Internet Archive

1. Datadog Pricing: Infrastructure, APM and Log Management plans, allotments and FAQ — Datadog

1. Custom Metrics Billing — Datadog Docs

1. Containers Billing — Datadog Docs

1. New Relic Compute pricing — New Relic

1. View and manage data retention — New Relic Documentation

1. Metrics pricing — Grafana Labs documentation

1. Logs pricing — Grafana Labs documentation

1. Traces pricing — Grafana Labs documentation

1. Host hours billing — Grafana Labs documentation

1. Getting Started with Datadog Sites — Datadog Docs

1. Grafana Cloud regional availability — Grafana Labs documentation