Datadog vs Grafana Cloud

At a glance

1. One sells a host, the other sells a series

Datadog's infrastructure line is a licence: $15 per host per month billed annually, $18 month-to-month, and every licence carries 100 custom metrics. Grafana Cloud's Pro plan is a $19 platform fee for the whole account, with 10,000 active series and 50 GB of logs inside it.

That is the whole philosophical difference, and it decides the invoice shape:

• Datadog: cost grows with machines first, then with the cardinality of what you send from them.
• Grafana Cloud: cost is flat until you pass the account-level allowances, then grows with data volume only.

Neither page hides this. It is written in the billing unit, which is the one line worth reading before the price.

2. The custom metrics line is the one that grows without a deploy

Per unit, Datadog's custom metrics cost $5 per 100 per month, which is $50 per 1,000. Grafana Cloud's first paid band is $6.50 per 1,000, then $5.90 above 100,000 and $5.50 above 200,000.

Datadog's rate is 7.7 times Grafana's, and the allotment it ships with is 100 times smaller. Those two numbers pull in opposite directions, so the bill depends on whether your cardinality scales with hosts or with users.

3. Two shops, and the same four numbers

Shop one: 20 hosts, 5,000 custom series.

Shop two: 20 hosts, 200,000 custom series, which is what a user_id tag does to a counter.

The second table is the one that arrives without a deploy, a migration or a new service. It arrives the week someone adds a tag.

4. Logs: one vendor counts bytes, the other counts events

This is where the two sheets stop being directly comparable. Datadog charges $0.10 per ingested GB, then $1.70 per million indexed events at 15-day retention or $2.50 at 30-day. Grafana charges three meters per GB: $0.05 to process, $0.40 to write and $0.10 to retain, with 50 GB included.

Take 500 GB a month at roughly 2.5 KB per event, or 200 million events, indexed for 30 days.

The event meter rewards fat logs and the byte meter rewards few of them. The same 500 GB is $550 or $247.50 depending on which unit your logging happens to satisfy.

5. Both vendors now discount you for deleting tags

Datadog sells Metrics without Limits, which splits one number into two: ingested custom metrics at $0.10 per 100, and indexed ones at $5 per 100. Grafana ships Adaptive Metrics, which aggregates and drops unused series, and the page states the claim as the vendor's own: up to 80%.

Read that as an admission. When a vendor builds a product whose only job is to delete time series before they reach the meter, the meter is cardinality, and the vendor knows the cardinality is often junk.

The practical move costs nothing:

1. Sort your metrics by tag value count, not by metric name.
2. Drop any tag whose values scale with users, carts or request ids.
3. Re-count before you negotiate a discount, because the discount is priced on the number you can already fix.

6. What the switch actually costs

Grafana Cloud speaks Prometheus for metrics, Loki for logs and Tempo for traces, so an application already instrumented with a Prometheus client can point at a remote write endpoint and stop. Datadog's agent, dashboards, monitors and APM traces do not translate: every dashboard is a rebuild, and every monitor needs a PromQL equivalent.

Retention also moves. Datadog keeps full-resolution metrics for 15 months on Pro; Grafana's Pro plan lists 13 months for metrics and 30 days for logs, traces, profiles and k6 tests.

The reverse move is the expensive direction for a small team: going to Datadog means adopting a per-host licence where Grafana charged per account, so the same twenty hosts start at $300 before a single custom metric.

Cost arithmetic at volume

These are computed from the published unit prices, not measured. Each row prints its own formula, so you can replace our assumption with your traffic and redo it.

Decision matrix

Rows are the situation you are actually in. Pick the row before you read the pricing page again.

Code and configuration

# Every tag value pair is its own billable series on Datadog.
from datadog import statsd

def checkout(cart_id: str, user_id: str, items: int) -> None:
    statsd.increment("checkout.completed", tags=[f"user_id:{user_id}"])
    statsd.gauge("checkout.items", items, tags=[f"cart_id:{cart_id}"])
# Breaks when: user_id has 50,000 values. That is 50,000 custom series,
# against an allowance of 100 per host — 500 hosts' worth of licence for one tag.

# Same job, same mistake: cardinality is the meter, not the vendor.
from prometheus_client import Counter, Gauge, start_http_server

checkout = Counter("checkout_completed_total", "Checkouts", ["user_id"])
items = Gauge("checkout_items", "Items per cart", ["cart_id"])
start_http_server(9100)

checkout.labels(user_id="u_123").inc()  # one time series per label value
# Breaks first here: 50,000 user_id values are 50,000 series on Grafana Cloud too,
# and the 10,000-series allowance ends long before the first page of users.

What you gain, what you give up

Migration path

Common questions

Sources & verification

Prices and limits were read from these pages on 2026-11-10. Vendor marketing claims are quoted as claims. Anything derived is arithmetic on the numbers below.

Keep reading