Salta al contenuto principale
Servizi

AEM as a Cloud Service migration: plan, code and content

From the Best Practices Analyzer to go-live: repository refactoring, Dispatcher, Content Transfer Tool and Cloud Manager pipelines managed as a single project.

An AEM as a Cloud Service migration is not a version upgrade. It is a change of operating model, and it must be planned as one. The code has to respect the split between immutable and mutable content, OSGi configurations must be managed per environment, Oak indexes must be delivered as code, the Dispatcher must pass SDK validation, and releases go exclusively through Cloud Manager. Those who approach the project as a simple port discover the problems late, when fixing them costs more.

Our approach to AEMaaCS migration is phased, with explicit exit criteria. It starts with an assessment based on the Best Practices Analyzer and manual analysis, continues with repository and code refactoring, sets up the new delivery process, and only then transfers content with the Content Transfer Tool, in several passes to shorten the editorial freeze window. Each phase ends with verifiable evidence.

The end-of-support date of the AEM version in use, and the contractual deadlines that follow from it, should be verified on official Adobe channels and in your own agreement [VERIFICARE]. We do not use third-party deadlines to justify a project: we start from the real state of the installation and the business goals, and build a calendar the team can sustain.

Pre-migration assessment with BPA

The Best Practices Analyzer, available through Cloud Acceleration Manager, analyzes an existing instance and flags patterns that need attention before the move: use of repository areas that can no longer be modified, incompatible configurations, unsupported APIs, workflows and replication agents to review. The report is an excellent starting point but does not set priorities: two findings with the same severity can have very different effort.

For this reason we pair the BPA with a manual review of the project, covering Maven modules, dependencies, /libs overlays, servlets registered on paths, integrations with external systems and scheduled jobs. The result is a refactoring plan with estimates and dependencies, in which open architectural decisions, such as what to retire instead of migrate, are highlighted and discussed with the client before starting.

Repository refactoring and project structure

On the cloud, code is deployed to /apps and /libs as immutable content, while /content, /conf, /var and /home remain mutable. Anything the code writes to immutable areas at runtime must be moved. The AEM Project Archetype structure is adopted, with separate modules for bundles, components, configurations, initial content and front-end, and a single container package that Cloud Manager can build reproducibly.

Tools such as Repository Modernizer help separate mutable from immutable content, while Index Converter and Dispatcher Converter automate part of the conversion of indexes and configurations. They should be treated as accelerators, not solutions: the output must always be reviewed. We pay particular attention to custom indexes, because a badly written or missing index in production turns into slow queries that nobody will notice until traffic grows.

OSGi configurations, Dispatcher and Cloud Manager

OSGi configurations are organized by run mode and environment, secrets and variable values go through Cloud Manager variables, and dependencies on the local instance must be removed. Development, stage and production environments should differ only in configuration values, never in code. The Dispatcher must be moved to an SDK-compatible structure, validated locally with the validation tool before every pipeline run.

In Cloud Manager we define separate pipelines for application code, front-end and web tier configuration, with quality gates on code quality, security and performance. A slow or unstable pipeline is as much a risk to the project as an application bug, so its design is part of the plan and is tested early, not the day before go-live.

Content transfer with the Content Transfer Tool

The Content Transfer Tool (CTT) extracts content from the source instance and loads it into the cloud environment. For large repositories we plan an initial full extraction and subsequent differential transfers (top-ups), so that the content freeze window is limited to the last pass. Test run results are recorded and compared, because real behavior depends on size, structure and data quality.

Before transferring it pays to clean up: obsolete versions, duplicate assets, test content and unused branches. After ingestion we verify permissions, references, pending workflows, URLs and redirects, and perform a sample editorial validation. Go-live includes the DNS switch, CDN configuration with custom domains and a rollback plan agreed with whoever manages the infrastructure.

After go-live

The period after the move is when differences from the previous environment emerge: different replication times, caching behavior, logs to be consulted in a new way, editorial habits to adapt. We plan a stabilization phase with log monitoring, cache hit checks and verification of pipelines under real release load, and we train the team on the new way of diagnosing problems.

From here the migration connects naturally to managed services: continuous platform updates require up-to-date dependencies, automated regression tests and a managed backlog. Timelines and costs of a migration depend on repository size, number of sites, integrations and the quality of the starting code, and should be estimated after an assessment.

FAQ

AEMaaCS migration FAQ

It depends on repository size, code quality, integrations and number of sites. We do not give generic durations: after an assessment with the Best Practices Analyzer and manual analysis we can estimate phases and calendar.

Not necessarily. It is often an opportunity to retire sites, versions and assets that are no longer used. The decision on what to transfer should be made before the CTT, because it affects transfer duration and storage costs.

It extracts content from the source instance and ingests it into the cloud environment. It is used in several passes, with differential transfers, to reduce the time during which editors cannot work. It does not replace code refactoring or the configuration of Dispatcher and pipelines.

Almost never without changes. The cloud requires a specific configuration structure and passing SDK validation. Dispatcher Converter helps, but the final configuration must be reviewed by hand, in particular filters, cache rules and rewrites.

The date should be verified on official Adobe channels and in your agreement [VERIFICARE]. We do not use it as a sales argument: we prefer to build the plan on the state of the installation and business goals.

Plan your AEM as a Cloud Service migration

Share your current version, hosting model and repository size. We will propose a pre-migration assessment with a clear scope and output.

Request a migration plan