Skip to main content

Stack: Databricks

Okta disaster recovery for teams running on Databricks

Most Databricks customers federate authentication through Okta — SCIM-provisioned workspace users, SSO-gated workspace access, and group-driven cluster permissions. If the Okta-side configuration drifts, the people who can debug your Lakehouse incidents may not be able to log in to do it. The cost is data freshness, pipeline backlog, and on-call burnout.

Butterfly captures versioned, encrypted snapshots of the Okta configuration that governs Databricks access — the SAML or OIDC app integration, the SCIM provisioning feed, the assigned groups, and the sign-on policy. Restore preview shows the exact diff before any revert.

What you get

How Butterfly fits Databricks

Snapshot the Databricks app integration

Every backup captures the Okta-side Databricks SAML or OIDC app: the attribute mapping, the assigned groups, the sign-on policy, and any custom claims that govern workspace selection.

SCIM feed for Databricks workspace users

The SCIM connection that provisions Databricks workspace users is part of every snapshot. If a teammate disables it during a cleanup, restore preview shows the configuration delta and the population that would be affected.

Group rules that drive cluster permissions

Group rules are how most teams scale who-can-run-which-cluster. Butterfly versions every rule. Restore preview tells you which Databricks-bound groups would gain or lose members before you commit.

What goes wrong

Three incidents you have already seen variations of

SCIM connection paused — new hires can't reach Databricks

A scheduled credential rotation paused the Databricks SCIM connection. New hires onboarded for the data platform team showed up in the org chart but not in the Databricks workspace. The gap surfaced as ticket volume the following week.

Sign-on policy change blocks a contractor population

A device-trust policy tightening was intended for full-time engineers. The group expression caught a contractor group that needed Databricks access for a customer-data project. Two days lost coordinating a fix.

Group rule deletion drops cluster admins

A group rule feeding the data-platform-admins group was deleted during a Friday-afternoon cleanup. Monday morning, no one could administrate the production clusters. Restore preview surfaces the rule and the membership delta.

Honest scope

What Butterfly captures — and what it does not

In scope

The Okta-side configuration governing Databricks access: the Databricks SAML / OIDC app integration, attribute mappings, assigned users and groups, group rules, sign-on policies, SCIM provisioning configuration, and Workflows automations.

Out of scope

We do not back up Databricks notebooks, jobs, cluster configurations, Unity Catalog grants, or any workspace-side state. Databricks-side recovery is owned by Databricks tooling and your data engineering team.

Plans

Free, Standard, or Business

Free

$0 / forever

  • 1 Okta connection
  • 7-day retention
  • 1 total backup
  • No credit card

Standard

$1 / user / month — $99 minimum

  • 2 Okta connections
  • 90-day retention
  • Restore preview + dry-run
  • Audit Pack PDF (framework-filterable)

Business

$2 / user / month — $299 minimum

  • Unlimited Okta connections
  • Unlimited retention
  • Continuity (warm standby)
  • Priority restore support

Pricing reference: /upgrade. Provider coverage today: Okta, Okta Workflows, Auth0.

Regulatory shape

Compliance and audit angle

SOC 2 CC6 (logical access), ISO 27001 A.5.16 (identity management), and HIPAA 164.312(a)(1) (access control — for healthcare data teams) all expect identity-layer continuity. Butterfly's Audit Pack PDF maps these.

Butterfly's own SOC 2 Type II work is in progress; current status lives in the Trust Center.

Frequently asked

FAQ

Does Butterfly back up Databricks notebooks or jobs?

No. Butterfly only backs up the Okta configuration that governs how your team reaches Databricks. Databricks-side artifacts are handled by Databricks' own tooling.

How are SCIM connections versioned?

Each backup captures the connection configuration, the assigned groups, and the attribute mapping. Restore preview shows the diff between any two snapshots.

How long are backups retained?

Free: 7 days. Standard: 90 days. Business: unlimited.

Recover your Okta org in minutes, not hours

Talk to Mick (the founder) for a 30-minute demo, or start the free trial. No credit card for the free tier.