Sandboxes, Deployment & Release Management

Sandbox types, change sets vs Metadata API / DevOps Center, and a safe release process.

Admin 2 min readSandboxesDeploymentChange setsDevOps

Sandbox types

TypeCopiesData storageRefresh interval
DeveloperMetadata only200 MB1 day
Developer ProMetadata only1 GB1 day
Partial CopyMetadata + sample data (sandbox template)5 GB5 days
FullMetadata + all dataSame as production29 days
  • Developer/Developer Pro for building; Partial for QA with realistic data; Full for UAT, performance and staging.
  • Scratch orgs (Salesforce DX) are short-lived, source-driven orgs created from a definition file.
  • After a refresh, user emails get a .invalid suffix so sandboxes don't email real users.

Moving changes between orgs

ToolNotes
Change setsPoint-and-click between connected orgs (production ↔ its sandboxes); no deletes; must add dependencies manually
DevOps CenterSalesforce's change-tracking UI with Git/GitHub behind it
Salesforce CLI / Metadata APIsf project deploy start; scriptable, works with Git and CI/CD
Unlocked / managed packagesModular, versioned releases

Deploying Apex to production

  • Code can't be edited directly in production.
  • Deployments run tests: ≥ 75% overall coverage, every trigger covered, all tests passing.
  • Test levels: NoTestRun (sandboxes only), RunSpecifiedTests, RunLocalTests, RunAllTestsInOrg.
  • Validate (check-only) a deployment ahead of time, then use Quick Deploy during the release window.

A safe release flow

  1. Build in a Developer sandbox (or scratch org), commit to Git.
  2. Deploy to an integration / QA sandbox; run automated tests.
  3. UAT in a Partial/Full sandbox with business users.
  4. Validate against production days before; fix test failures.
  5. Quick Deploy in the release window; run smoke tests; communicate changes.

Interview questions

  1. Which sandbox would you use for UAT with real data volumes? → Full (or Partial Copy if data needs are smaller).
  2. Limitations of change sets? → Connected orgs only, no destructive changes, manual dependency handling, no version control.
  3. What's needed to deploy Apex to production? → 75% coverage, triggers covered, passing tests.