PLATFORM · OBSERVABILITY

You'll know before
the customer does.

Logs, traces, metrics, and exceptions — first-class, queryable, retained 30 days. No agent to install, no Datadog bill to argue, no second vendor to audit. Open the dashboard, type a question, get an answer.

Try Scalable → Read the docs
30dretention by default 4pillars, one query 120msp95 query latency 0agents to install
01Four pillars

Logs, traces, metrics, exceptions.
One query bar.

Cross-correlate without leaving the page. Click a trace span → see the logs. Pin a metric → get the spans behind the spike. Group exceptions by region.

01 · LOGS

Structured, indexed.

JSON-native. Pin any field as a dimension. Search 30 days in <120ms. Cardinality is yours, not ours.

  • retention30d default · 1y max
  • indexingauto, all fields
  • query p95118ms
14:02:11 200 POST /checkout 87ms
14:02:11 200 GET /cart 14ms
14:02:12 500 POST /webhook · panic
02 · TRACES

Distributed, untruncated.

OpenTelemetry-native. Auto-instrumented for 38 frameworks. Every span retained — no head sampling, no surprises.

  • samplingtail-based, 100%
  • compatibilityOTLP, Jaeger, Zipkin
  • span-to-log jump1 click
03 · METRICS

Sub-second granularity.

RED, USE, custom — Prometheus-compatible. 1-second points for 7 days, 1-minute for 30. Histograms, not averages.

  • resolution1s for 7d
  • query langPromQL + SQL
  • histogramsnative
04 · EXCEPTIONS

Grouped, deduped.

Stack-trace fingerprinted. New regressions surface within 30s. Linked to the deploy and commit that introduced them.

  • fingerprintingframe-aware
  • new-issue alert30s median
  • deploy linkageauto
NEW · PaymentTokenExpired
checkout-svc · introduced in 3a8f10d
02Query in plain English

Ask in English. Answer in 120ms.

Type a question. We compile it to PromQL, SQL, or trace-DSL behind the scenes. Or write the query yourself — we don't gatekeep.

Ask

"Which endpoints regressed after the 18:42 deploy?"

# compiled to ↓ SELECT endpoint, quantile(latency_ms, 0.95) AS p95, count(*) AS rps FROM traces WHERE service = "checkout-svc" AND ts BETWEEN "18:42" AND now() GROUP BY endpoint HAVING p95 > prev(p95) * 1.5 ORDER BY p95 DESC
3 endpoints regressed · queried 432M spans in 118ms

POST /api/checkout   p95 87ms → 318ms · +266%
GET  /api/cart      p95 14ms → 38ms  · +171%
POST /api/webhook    p95 22ms → 41ms  · +86%
Ask

"Show me only errors from EU users."

# compiled to ↓ FROM logs WHERE level IN ("error", "fatal") AND user.region IN ("eu-central", "eu-west") AND ts > now() - 15m GROUP BY exception.fingerprint ORDER BY count DESC
2 distinct exceptions · 47 events · 118ms

PaymentTokenExpired  ·  41 events · checkout-svc
RedisConnectionPool  ·  6 events  · session-svc

→ click any row to see the spans, logs, and the deploy that introduced it.
03Stream out, too

Already on Datadog? Stay there.

Scalable observability is the default — open out of the box, no contract — but if you've already invested in another stack, we'll mirror everything to it. Two-way OTLP.

Datadog Grafana Cloud Honeycomb New Relic Splunk OpenTelemetry Collector S3 / Parquet BigQuery Snowflake Sentry PagerDuty Slack & Discord
"We deleted the Datadog contract. Saved $48k a year and stopped fighting over cardinality limits."
AV
Alex VolkovHead of Infra · Ridgeline · 42 services on Scalable
Annual obs bill$48k → $0
Time to root-cause21m → 4m
Customers who reported it firstnone.

See your stack before
your users see it crash.

Built-in for every service on Scalable. No agent, no contract, no fight over indexing. Just open the dashboard.

Try Scalable → Read the docs