How to read the observability cost comparison

Illustrative monthly estimates using vendor rates. Retention is modelled at 90 days where published pricing supports it; otherwise the vendor's standard retention applies. Retention and included costs differ. Groundcover includes the lower BYOC hosting estimate. AWS AMP is metrics only. Volume and negotiated discounts are excluded.

Last updated: .

How is the xScaler estimate calculated?

Monthly USD = $19.00 + active series × $0.002175 + log GB × $0.45 + trace GB × $0.45.

100 servers · 1 TB logs · 100K metric series · 300 GB traces · a month: $19.00 base + $217.50 metrics + $450.00 logs + $135.00 traces = $821.50 per month.

This is a paid Scale estimate with 90-day retention. All entered usage is charged at the displayed Scale rates. Free plan allowances are not deducted. Host count does not change the xScaler estimate. Confirm support, tax and contract terms before purchasing.

Model your workload in the cost calculator

How are vendor bills calculated?

Each estimate uses the relevant workload inputs, vendor rates, allowances and billing units. The calculator shows vendor totals and an itemised xScaler Scale estimate. The assumptions and source links below explain how the competitor estimates are modelled. Only relevant inputs affect each estimate. Host count does not change xScaler's charge, and AWS AMP covers metrics only.

Vendor-rate estimates for the same inputs. Retention and included costs differ as shown below.
OptionMonthly USDBasis
AWS AMP$262.02Metrics only · 90 days · excludes label metadata and managed collection
xScaler$821.50Scale plan · 90 days for all signals
New Relic$989.20Data Plus · 120-day logs / 98-day traces · 1 user
SigNoz$1,298.4090-day logs/traces · 3-month metrics · Teams usage minimum
Self-hosted$1,405.00AWS infrastructure estimate · 90 days · excludes engineering labour
Grafana Cloud$1,409.0090-day logs/traces · 13-month metrics · before volume discounts
Dash0$1,644.0030-day logs/traces · 13-month metrics · before volume discounts
Groundcover$5,314.75Enterprise + BYOC · 90-day retention
Dynatrace$9,031.7290-day logs/traces · 462-day metrics · DPS commitment
Datadog$12,448.0015-day indexed logs/spans · monthly Infrastructure Pro + APM

How should retention and scope be read?

  • The model assumes a 60-second metric sampling interval and a 30-day month (720 hours). Volumes use decimal units: 1 TB = 1,000 GB and 1 GB = 1,000,000,000 bytes. Vendor GiB-based rates use 1 GiB = 1,073,741,824 bytes.
  • Zero inputs remove the corresponding usage charges. AWS AMP query charges are zero without metrics, and Dynatrace log/trace query charges are zero without either signal. Host-priced models retain infrastructure licences for the entered hosts. Groundcover also includes its estimator's host-generated telemetry. Service minimums, access fees and self-hosted infrastructure remain, so a single-signal workload is not a standalone-product quote.
  • USD, US deployment, continuous hosts and scalar series. All records are ingested without filtering. Assumes 1 KB per log/span, 100% custom metrics and APM hosts, 100 assumed bytes per New Relic metric sample, one access user, 1 TB log/trace queries and one billion AMP query samples per month. Features and support differ.
  • Retention is modelled at 90 days where published prices allow adjustment. Datadog uses 15-day indexed logs/spans, and Dash0 uses 30-day logs/traces with 13-month metrics. These are price comparisons, not equivalent-service quotations.
  • Grafana Cloud uses published processing and writing rates, the 50 GB written allowance per signal, and two additional 30-day retention increments. Processing and retention do not subtract the written allowance. Automatic volume discounts are excluded.
  • Dynatrace assumes eligible, evenly distributed host-origin scalar metrics and applies 1,500 included points per host per 15 minutes. Metrics include 462 days. Ten trace days are included, with 80 additional days charged. Retained bytes are assumed equal to ingested bytes before compression.
  • New Relic Data Plus includes 120-day logs and 98-day distributed traces. Dimensional metrics have 30-day raw retention and 13-month one-minute rollups. For this estimate, we assume 100 billable bytes per metric sample. This conversion has not been verified against New Relic billing. It includes one Standard full-platform user. The 100 GB ingest allowance applies once across signals.
  • Groundcover uses Enterprise licence rates plus its public US East (N. Virginia) hosting estimator at 90-day retention. The lower hosting estimate is plotted. The estimator adds 10,000 metric series and 4 GB/day of traces per host to the custom telemetry inputs.
  • AWS AMP is a metrics-only component estimate, excluding label metadata and managed collection. Its bar has no savings percentage or multiple. Self-hosted is a three-AZ AWS infrastructure planning estimate for Mimir, Loki, Tempo, Grafana and alerting, with a separate Kafka pool. It excludes engineering labour and is not a benchmarked capacity guarantee.
  • Prices exclude tax, volume and negotiated discounts, migration, operational labour and unselected add-ons. Feature, support and SLA equivalence is not asserted.
  • Self-hosted sizing: minimum three m6i.2xlarge workers and three m6i.large Kafka brokers on EKS in us-east-1. Worker groups scale with combined series/500,000 + daily log GB/100 + daily trace GB/50, rounded up to whole three-node groups. These thresholds assume moderate query load and reserved headroom, not measured throughput. Brokers scale per 500 GB/day in three-node groups. Host count alone adds no charge.
  • Self-hosted storage: 90-day steady state, 2 stored bytes per metric sample plus 50% metadata, logs compressed to 20% and traces to 30% of input. Kafka assumes 16 wire bytes per metric sample, three replicas and 24-hour retention with 50% disk headroom. Each worker and broker has a 100 GiB gp3 minimum. S3 is not multiplied by replica count. Compression, series churn and query workload need validation for an actual deployment.
  • Self-hosted networking assumes two cross-zone replica transfers per ingested byte, 1 TB/month query scans with two-thirds cross-zone, 100 GB/month internet query results, one ALB, three NAT gateways with 100 GB/month maintenance traffic, and six public IPv4 addresses. S3 uses gateway endpoints. S3 requests assume 4 MiB objects and three writes per stored object. gp3 uses included performance and S3 uses its starting storage rate. Connection-driven ALB LCUs, extra IOPS, DR copies, paid AWS support, tax, account credits and shared free allowances are not modelled. Stack monitoring runs on the worker pool.

Pricing sources and model assumptions

How should you compare the options?

  1. Measure a representative month of active series, scrape interval, logs, traces and query activity.
  2. Agree the same retention, region, features, support and export requirements for each option.
  3. Calculate every required product and allowance from a dated rate card or written quote.
  4. Run a representative workload alongside the existing platform and verify the actual usage and bill.

An existing provider may be the better fit when its integrations, broader product coverage, support commitments or migration cost outweigh a modelled rate difference. Self-hosting may suit a team with spare operating capacity and specific control requirements. Evaluate these differences before making a purchasing decision.

Datadog comparison · Grafana Cloud comparison

Compare against your actual bill.

Bring a recent bill and your workload requirements. We will review retention, capabilities, support and relevant contract terms, then agree what to validate in a month-long parallel PoC. Extensions are available by agreement.

Compare my costs