Compare true total cost of ownership — infrastructure and human operational costs — across observability platforms. Or open Serverless Estimator for Elastic Observability Complete (logs, metrics, traces) like the Cloud pricing form.
Estimation purposes only
This tool is not validated by Elastic Product, Marketing, or Competitive Intelligence for accuracy. List rates reconciled as of August 2026 against the Elastic Cloud Serverless and Cloud Hosted pricing tables plus ad hoc field feedback (e.g. Serverless estimator alignment). See also Serverless Observability pricing (TSDS metrics effective July 1, 2026). Other vendor rates and unit conversions are approximate. Confirm all figures with official pricing, contracts, and measured usage before customer-facing quotes.
Schemaless data blocks: Observability is schemaless at ingest — Streams uses AI to identify patterns, so there is no custom-metric SKU; the billable path is ingested GB/TiB (plus retention), not per-series metric packs. One block can carry mixed OTLP. Open the Data Blocks tab for illustrative $/TiB-month at 1, 50, 100, and 500 TiB/mo grounded in Serverless Observability pricing Complete floors ($0.09/GB ingest · $0.019/GB-mo retention for logs/traces; TSDS metrics $0.023 / $0.005) plus Streams → S3 object storage. Per-tab comparisons may still apply signal-specific metering for vendor parity (TSDS 25%, logs 1.66×, etc.).
Not included: Datadog/Splunk/etc. use approximate public list pricing; Grafana Cloud Metrics uses Pro billable series (samples/sec × 60 DPM ÷ 1 included DPM, $6.50/$5.90/$5.50 per 1k after 10k included) — not per million datapoints. Elastic is GB volume; Grafana/Datadog are unique-series meters, so high cardinality can favor Elastic and fat/low-cardinality samples can favor them. Operational FTE is optional. Excludes synthetics, LLM observability, Agent Builder, support ECUs, Cross-project Search mounted-GB charges, Adaptive Metrics, and negotiated discounts.
Compare Elastic Cloud Hosted and Serverless with Datadog host and custom-metric pricing. The comparison chart and platform list update to match.
EDOT is Elastic's OTel distribution (merging into contrib). Agent/Beats metrics use the OTel/EDOT size.
Each tag multiplies your metric volume by this number
Datadog bills Infrastructure Pro ($15/host/mo) on Metrics plus custom metrics ($0.05/series/mo after included series), and APM Pro ($31/host/mo) on Tracing. Host count is separate from samples/sec.
Estimated 250 hosts (linux/windows/k8s-node inventory, or log GB/day ÷ 0.04 GB/host/day). Prefer manual hosts in Samples/sec mode.
Serverless uses 1-day hot · Streams → S3: 1-day hot on Observability Complete, then Streams sends the rest to object storage (S3). ECH uses the same total retention window as 1-day hot + ILM writable-frozen blob (Cloud Hosted pricing). Self-hosted stays a fixed infra estimate.
In short: keep recent data searchable for 1 day, then age the rest to object storage for long-term retention. Serverless ages to S3 (~$0.023/GB-mo); ECH ages to ILM blob (~$0.033/GB-mo snapshot) for the same total window. Metrics ingest + Serverless hot use the Observability Complete volume tier table × 25% (TSDS). ECH also includes the ~$200/mo cluster minimum.
All ECH rows use 1-day hot · ILM → blob (writable frozen) — 1-day data hot (RAM-hour) plus ILM on blob (writable frozen) for the same total retention as the slider above. Full-fidelity ingest; no Streams sampling on ECH.
Serverless uses 1-day hot · Streams → S3 plus the ingest shaping below. Dial each lever’s % volume removed to match how aggressive this customer wants to be — every environment differs. Chart totals update as you change these.
Drop unused / noisy metric series before ingest.
Roll up high-cardinality series (TSDS / stream processors).
Reduce resolution on data older than the 1-day hot window before/while it lands on S3. Hot stays full fidelity.
Sample stream policies for this signal — collapsed
Self-hosted platforms need engineering time to run. Fully managed SaaS (Elastic Serverless, Grafana Cloud, Datadog, etc.) is $0 platform ops. Enable only when comparing DIY Prom/Mimir/Thanos-style stacks.
Egress costs excluded from calculations. Enable to see full TCO including data transfer costs.
SSE-KMS · bucket policies · lifecycle to Glacier optional
Analysts query cold archives and partner buckets in place. No full re-ingest into hot tier unless you choose to promote data via a connector.
ILM searchable snapshots, compliance exports, and connector sync jobs write structured objects back to S3 for durability and cross-region DR.
S3 storage + API requests are billed by AWS; Elastic charges ingest/retention on promoted or hot-tier data. Federated read avoids duplicating bytes on hot RAM.
Illustrative architecture for customer conversations. See Cross-project search, S3 snapshot repository, and Elastic connector docs for your deployment type.
Internal estimates only — not validated by Elastic Product, Marketing, or Competitive Intelligence. Confirm with official pricing and customer usage before external quotes.
Elastic bills total ingested volume (~77.2 GB/mo at 320 B/sample). Datadog custom metrics are also a unique series meter (plus hosts), not GB. High cardinality or faster scrapes inflate series bills while Elastic stays on GB — often a large Elastic advantage. Fat samples / few series can flip it the other way. Move samples/sec and bytes/sample to see both sides.
Bar lengths use a log scale because costs span more than 20× ($11 → $3.8K). Dollar labels stay linear.
Pricing is based on publicly available list rates as of August 2026 (AWS us-east-1 Serverless and Cloud Hosted pricing tables). Elastic Observability Serverless Complete and TSDS metrics rates per elastic.co pricing (TSDS metrics effective July 1, 2026). All figures are estimates — actual costs vary with negotiated discounts, committed use, and deployment configuration. Use the Full Stack TCO tab for a complete cross-signal comparison. Contact your SA for a custom TCO analysis.