Encryption
Every connection to Sazabi is encrypted with TLS 1.2 or higher in transit, and data at rest is encrypted with AES-256. Telemetry is protected from your infrastructure to Sazabi’s storage layer, and the credentials you connect are encrypted before they are stored.Storage and retention
Sazabi uses managed log storage — there is no storage backend for you to provision or secure. Everything you send is indexed for the agent’s queries and kept for your plan’s retention window. When data ages past that window, it is deleted automatically; logs older than the window are not retained. See Log storage & retention for the full model. If your organization needs logs processed in its own cloud account and stored in infrastructure it owns, see Bring your own cloud (BYOC).Data residency
You choose the region your project’s data lives in when you create the project — a United States or European Union region — so telemetry stays in the region that matches your data sovereignty requirements. The region is fixed for the life of the project. See Log storage & retention for how to choose a region.Deletion
Your data leaves Sazabi in three ways:- Retention expiry — data is deleted automatically once it passes your retention window.
- Stream or source removal — removing a log source stops new data; historical records age out on the normal retention schedule.
- Project deletion — deleting a project removes the telemetry stored for it.
How your data is used
Sazabi is a data processor: it works with your data only to run the product for you. Your data is never used to train models — neither Sazabi’s nor its providers’. Sazabi uses enterprise API tiers that contractually exclude submitted content from model training, and passes that same commitment through to you in its Data Processing Addendum. No model is tuned on your data without an explicit, written opt-in.Redact sensitive data
You decide what leaves your systems, and Sazabi can redact common sensitive values for you at intake.Redact at intake
Turn on Redact sensitive data in Settings > Project to redact logs for every log stream in the project. Sazabi replaces matching values with a placeholder before ingested logs are stored or shown to the agent:
Values under attribute names that always carry a secret, such as
authorization, cookie, or password, are redacted whatever they look like. A redacted log records the kinds that were replaced in its sazabi.redactions attribute.
- The setting is off by default. Project admins can change it; there is no CLI equivalent.
- It applies to logs received after you turn it on. Logs already stored are not changed.
- Requests that fail intake, such as a malformed or oversized payload, can be kept unredacted in Sazabi’s internal failure queue.
- Detection uses fixed patterns, so it does not catch names, addresses, health information, or other free-form personal data.
Redact at the source
For data that must never reach Sazabi, strip it at the source — redact secrets and sensitive fields at your log source or collector, the point where you own the data. The Production checklist walks through this decision before you rely on Sazabi in production.Further reading
Log storage & retention
Managed storage, the retention window, and choosing your region.
Compliance
Certifications, audit reports under NDA, the DPA, and subprocessors.