The observability console your on-call deserves: live tail, queries, APM, alarms and dashboards on one timeline, card data masked before it is stored, and a bill you can see coming. Readers are always free.
| Endpoint | rpm | p99 | Errors |
|---|---|---|---|
| POST /v1/checkout | 612 | 182 ms | 0.4% |
| GET /v1/orders/:id | 420 | 38 ms | 0.0% |
| POST /v1/refunds | 96 | 240 ms | 2.1% |
Speaks OpenTelemetry. Works with the stack you already run.
The thread
Four tabs, three vendors and a spreadsheet is not an incident process. In Tracery a trace opens with its logs, an error links to the span that threw it, and an APM chart drills down to the requests behind the spike.
p99 on POST /v1/checkout crosses 800 ms. Slack, Discord, email or your webhook gets it — once, on the transition.
The payment dependency is the one turning red. One click opens example traces from that exact minute.
The waterfall puts the slow span on screen, with every log line it wrote right below it.
The issue is marked as a regression since the last deploy, with its stack trace and an owner.
Product
One console for the whole journey: from the first log line to the dashboard your CTO opens on Monday.
Log groups and streams created on first write. Live tail streams every matching event with one update per second. Filter by text, JSON fields, space-delimited columns or regex, then share the exact view as a URL.
Fields, filters, stats, sort and bins with a histogram and charts. Saved queries in folders, history and CSV export.
Traces & service map
Search spans by service, status or duration. Open the waterfall and the logs of that trace are already there. The service map shows who calls whom, and how it is going.
Rate, errors and duration per service, endpoint, dependency and database query, derived from your spans. p50 to p99, availability, faults and throttles.
Exceptions from logs and spans grouped by fingerprint, with regression detection, owners, notes and the trace one click away. Get notified on new issues and regressions.
Turn log patterns into metrics, combine them with metric math and search, and alarm on anything. Composite alarms with suppressors. Actions fire only on state transitions, to email, Slack, Discord or any webhook.
A 24-column grid with metric, log, text and alarm widgets. Auto refresh, period override and “Add to dashboard” from anywhere.
Organizations, projects and roles, access per project, invites by email or link, two-factor authentication and an audit log of every change. Read-only members never cost a seat.
Get started
Tracery speaks OTLP over HTTP for logs, metrics and traces. Point the OpenTelemetry SDK or collector you already use at it and you are done.
Each project keeps its own logs, metrics, traces, alarms, dashboards and keys.
Scoped to one project, shown only once, stored only as a hash.
It shows up in the console in under five seconds.
Security & privacy
Payment-grade by default. The PAN guard validates every candidate with Luhn and masks it on the server before anything is written, in logs and in spans. On for every project, from the first event.
Every row carries its tenant and every read is filtered by the database itself. No tenant set means no rows, never all rows.
TOTP two-factor authentication, session list with revoke, and every session revoked on a password reset.
Owner, admin, billing and member roles, with full, read-only or logs-only access per project.
Who invited, promoted, rotated a key or changed retention, when and from where.
Ingest keys are scoped to one project, carry a checksum, are shown once and stored only as an HMAC.
From 1 day to 10 years per log group, with deletion protection. Lowering retention hides old data immediately.
Cost control
Observability bills are famous for one thing. We built the opposite: you see the cost before you act, while the month runs, and you decide where it stops.
Cost and forecast by item, project, log group and service, a daily chart, and the exact invoice preview next to it, taxes included.
Inviting, promoting or granting full access shows the seat impact. Changing retention shows the storage estimate.
A monthly budget with alerts on actual and forecast spend, by email and through your notification channels.
An organization spend cap and a daily ingestion cap per project, against log storms. Ingestion stops at the limit; reading, the console and alarms keep working.
Pricing
A small base, generous allowances and transparent overage. Usage is measured by the hour and billed on the 1st. Read-only and billing users are free on every plan.
For side projects and kicking the tires.
Hard limits, never a bill: ingestion pauses at 5 GB until the 1st, with a heads-up at 80%.
For teams running production.
Upgrade in the console with a card. First month prorated. Cancel anytime: Pro stays until the end of the cycle.
For scale, procurement and finance.
Same engine, same console, same controls — with a contract that fits your company.
Who is a seat? Owners, admins and members with full access to at least one project. Seats are measured hourly: someone who joins on the 16th costs half a month. Read-only members and billing users never count.
+ unlimited read-only and billing users: $0
Start free, upgrade when you need itEstimate in USD before taxes. Storage assumes steady state and ~7× compression (logs usually compress 5–10×). The console shows the exact invoice preview.
Our promises
We acknowledge a batch only after the write commits. Retries are deduplicated, so a resend never doubles your data.
Support, product, finance: give everyone read-only access and pay nothing for it.
When you set a spend cap, ingestion stops there. You chose to stop, so we never bill past it.
New prices apply to new subscriptions. Existing ones only move with an explicit migration and notice.
Seven days of normal ingestion to fix a payment. Reading, the console and alarms keep working the whole time.
OpenTelemetry in, CSV out. Your instrumentation works with any OTLP backend.
FAQ
Anything else? Write to us and a human who built the product answers.
Contact usNothing is billed. The Free plan has hard limits: you get a warning at 80% and ingestion pauses at 5 GB until the 1st of the next month. Reading, the console and alarms keep working. Upgrade to Pro anytime to lift the limit.
Owners, admins and members with full access to at least one active project. Read-only members, logs-only members and billing users are free and unlimited. Seats are measured every hour, so you pay the monthly average, not the peak.
Compressed, in GB-months: the average stored over the month. Logs usually compress 5–10×, so $0.30 per compressed GB-month is roughly $0.03–0.06 per raw GB. Lowering retention reduces storage right away.
No. Tracery accepts OTLP over HTTP (protobuf or JSON, gzip) for logs, metrics and traces. Any OpenTelemetry SDK or collector works, in any language.
Set a monthly budget with alerts on actual and forecast spend, an organization spend cap that stops ingestion at the limit, and a daily ingestion cap per project. The Cost & usage page shows month to date, forecast and the exact invoice preview.
Pro is paid by card in the console. Every subscription renews on the 1st (UTC) and the first month is prorated. Usage is reported hourly. Prices are in USD; applicable taxes are added at checkout. Business can pay by invoice.
You get emails and in-app banners, and ingestion keeps running normally for 7 days. After that it is throttled for 7 more days, then paused. Reading never stops, and paying brings everything back at once.
Yes, from the billing portal. Pro stays active until the end of the cycle and then the organization moves to Free.
Create a project, send a log and watch it land in under five seconds. Free forever, no card needed.
