SOFTWARE / SYSTEMS / AIEngineering news. Technical depth.
NEWS / Cloud · 2 MIN READ

GitHub previews live migrations from GHES to data-resident cloud

Enterprise Live Migrations continuously synchronizes active repositories before a short cutover, targeting large repositories that cannot tolerate long freezes.

Announcement: · From GitHub

GitHub opened Enterprise Live Migrations, or ELM, as a public preview on May 7. The service continuously synchronizes repository data from GitHub Enterprise Server to GitHub Enterprise Cloud with data residency, then drains remaining changes during cutover. The announcement positions it for active repositories and large monorepos where an extended freeze is difficult.

Synchronization changes the migration plan

Traditional export-and-import migrations put most data movement inside a maintenance window. ELM moves that work earlier while contributors continue using the source. GitHub says resource-level progress tracking exposes failures before cutover, and describes ELM as complementary to GitHub Enterprise Importer rather than its replacement.

The preview runs as a service on the GHES appliance and is driven by the elm CLI. GitHub listed specific minimum patch versions across GHES 3.17 through 3.20. Those prerequisites are part of the migration project: an organization on an older patch must first establish that upgrade path.

Rehearse identity and the final switch

A shorter code freeze does not remove migration risk. Teams should map users, teams, repository visibility, branch protections, apps, webhooks, Actions dependencies, and external URLs before syncing data. Rehearse representative repositories that contain large histories, pull requests, issues, releases, and Git LFS objects, then compare counts and permissions at the destination.

The cutover runbook should name who stops writes, how in-flight automation is drained, when DNS or integration endpoints change, and what evidence permits reopening work. Preview status also warrants a fallback: record the last safe decision point and how contributors will return to the source if validation fails. Continuous copying buys time for verification; it does not make the destination correct without an explicit acceptance check.

SOURCES & CONTEXT

See the original announcement for availability and release details.