Cloud migration in Morocco: 10 decisions before moving your applications
A decision framework to map dependencies, choose a strategy per application, and prepare testing, cutover, rollback, and cloud operations.
Cloud migration in Morocco is not a server-copying exercise. It moves a business service together with its applications, data, interfaces, identities, controls, procedures, and operating responsibilities. Before cutover, the organization should be able to explain why each workload is moving, what depends on it, how it will be tested, who owns the go/no-go decision, and how the previous state can be restored.
The short answer is simple: do not migrate an application until ten decisions are documented and tested. They cover the business outcome, scope, technical strategy, target foundation, data, security, testing, cutover, operations, and reversibility. This framework separates migration speed from risk control.
What is application cloud migration?
Application cloud migration is the controlled transfer of an application and its dependencies from an existing environment to a cloud platform, or from one cloud to another. Its scope can include code, databases, files, message queues, certificates, scheduled jobs, network connections, logs, and operational tooling.
This definition is more useful than “server migration.” An application may start after a move and still be unusable because a payment interface, directory, business printer, or network rule did not follow. Decisions therefore belong to the business service and its dependency chain, not only to a virtual machine.
Morocco-specific checks before choosing a target
Data location is not governed by one universal rule. Morocco’s Law 09-08 regulates personal-data processing. Articles 43 and 44 address transfers to a foreign country, and the CNDP describes the applicable cases and formalities. An organization should identify personal data, destinations, processors, and legal mechanisms before selecting a cloud region. The official CNDP guidance on foreign transfers also states that the underlying processing must have completed the appropriate formality.
Decree 2-24-921 specifically addresses entities and vital infrastructures that hold sensitive information systems or sensitive data. For that scope, the use of qualified cloud providers is regulated by the DGSSI. It would be inaccurate to apply this regime automatically to every SME; first determine the entity, system, and data classification. The DGSSI lists the regulated cloud framework and current status.
These legal and sector checks do not replace technical assessment. They shape it: location, encryption, administrative access, subcontracting, log retention, export, and continuity must appear in the architecture and contract. Seek specialist legal advice when classification or transfer conditions remain uncertain.
Decision 1 — Which business outcome justifies the move?
Start with an observable outcome: retire end-of-life equipment, make deployments reproducible, improve recovery, support expansion, or reduce a specific operating risk. “Move to cloud” is not an outcome. It cannot guide spending or determine whether a workload should remain on premises.
Name a business owner and a technical owner. Define acceptance criteria, acceptable downtime, functions that must not regress, and evidence required after cutover. A workload without an accountable decision maker should not enter a migration wave.
Decision 2 — What is the real scope and what depends on it?
The inventory should link applications, components, data, and people. For each workload, record users, critical periods, versions, licenses, operating systems, databases, volumes, inbound and outbound flows, certificates, service accounts, scheduled jobs, DNS, backups, and monitoring tools.
The dependency map must distinguish verified facts from open assumptions. Combine interviews, configuration review, network logs, and production observation. A dependency absent during a quiet week may appear at month-end. Keep the inventory versioned and validated by business teams.
Decision 3 — Which workloads should move, stay, change, or retire?
Choose a strategy for each workload. AWS Prescriptive Guidance defines seven common families: retire, retain, rehost, relocate, repurchase, replatform, and refactor. The vocabulary makes trade-offs explicit without requiring AWS as the destination.
- Retire: decommission a workload with no continuing value or active retention requirement.
- Retain: keep a workload temporarily because of hardware, license, latency, compliance, or unresolved dependencies.
- Rehost: move with minimal change to reduce lead time while accepting much of the existing architecture.
- Replatform: change selected components, such as adopting a managed database, without rewriting the entire product.
- Refactor: redesign a significant part of the application when the expected outcome justifies cost and change risk.
- Repurchase: replace it with SaaS or another product after analyzing data, integrations, and exit conditions.
- Relocate: move a compatible platform with minimal transformation.
A sound portfolio often uses several strategies. A single pattern may speed up wave one, but it can also carry forward technical debt, unnecessary licenses, and unchanged operating costs.
Decision 4 — Is the cloud foundation ready?
Before onboarding a workload, define accounts or subscriptions, environments, identity, privilege, connectivity, logs, backup, budgets, and provisioning rules. Microsoft’s Cloud Adoption Framework organizes adoption across strategy, plan, ready, adopt, govern, secure, and manage, reinforcing that the operating target should precede workload movement.
A cloud landing zone in Morocco makes these decisions repeatable. Keep it proportional: complete enough to provide identity, networking, evidence, and cost ownership, but not overloaded with controls nobody can maintain.
Decision 5 — How will data be copied and reconciled?
For every dataset, define the source of truth, volume, change rate, quality, encryption, destination, copy duration, and acceptable interruption. Choose among a full stop, continuous replication, temporary dual write, or another tested mechanism. Use representative volumes during rehearsal.
Define reconciliation controls: counts, totals, hashes, constraints, business samples, and critical transactions. A completed copy without a reconciliation report is not migration evidence. For location, transfers, and processors, use Kanteek’s guide to data hosting in Morocco and cloud region selection.
Decision 6 — Which identities and secrets exist before, during, and after?
Migration often creates a period when two environments coexist. Inventory human and service identities, require strong authentication, minimize privilege, separate preparation from approval, and document emergency access. Keys, certificates, and secrets should not be copied into scripts or tickets.
Clarify the shared-responsibility model. The provider secures selected layers, while the organization remains responsible for identities, configurations, data, applications, and use appropriate to the chosen service. Enable administration and access logs before testing so evidence exists from the first operation.
Decision 7 — What test evidence authorizes production?
A technical check that “the application responds” is insufficient. The matrix should cover business journeys, interfaces, permissions, performance, availability, backup, restoration, monitoring, security, and operations. Test datasets must follow protection rules; production data should not be copied without necessity and control.
Perform at least one real restore instead of merely confirming that a backup exists. Validate alerts, dashboards, and incident routing. Kanteek’s cloud disaster recovery plan for Morocco explains the difference between stated objectives and evidence obtained through exercises.
Decision 8 — How will cutover and rollback work?
The cutover runbook lists each step, owner, proof, maximum duration, and point of no return. It covers change freeze, final synchronization, reconciliation, DNS or routing change, smoke tests, communication, user opening, and heightened monitoring.
Rollback is not the phrase “restore the backup.” It specifies which environment becomes authoritative, how post-cutover writes will be reconciled, who decides, and within what time. Rehearse the runbook on a pilot wave. A useful go/no-go decision relies on pre-agreed evidence, not a feeling during a night window.
Decision 9 — Who operates the application the next morning?
Migration does not end at the first green dashboard. Define support hours, escalation, service objectives, incident procedures, dashboards, patching, vulnerability handling, backup, restore, and capacity management. Teams receive documentation and practice essential actions before cutover.
Set an observation period with exit criteria. Separate migration defects from historical behavior and future optimization. This prevents hypercare from becoming an open-ended project.
Decision 10 — How will cost, value, and reversibility be measured?
Build a cost model before migration: compute, storage, backups, logs, network traffic, licenses, support, and operations. Attach every resource to an owner, product, and environment. The FinOps in Morocco guide explains invoice reconciliation, allocation, and anomaly management without inventing savings percentages.
Reversibility completes the model. Document export formats, volumes, required tools, expected duration, egress costs, proprietary dependencies, and backup treatment after termination. Run a meaningful export test. A contractual exit clause without a tested technical path remains an intention.
What deliverables should be required before go/no-go?
- a versioned inventory with owners and criticality;
- a validated dependency and data-flow map;
- a reasoned migration strategy for every workload;
- a target architecture and landing-zone rules;
- a data plan covering transfer, reconciliation, and protection;
- an access, responsibility, and logging matrix;
- business, performance, security, and restore test reports;
- a wave plan and rehearsed cutover and rollback runbooks;
- a cost model, assigned owners, and alerts;
- an operating model and tested export evidence for reversibility.
Common cloud migration mistakes
- selecting a provider before understanding the application portfolio;
- using one migration strategy for every workload;
- missing scheduled jobs, certificates, and low-frequency interfaces;
- treating a completed data copy as automatically correct;
- testing availability without testing restoration;
- planning cutover without an accountable go/no-go owner;
- ignoring network, logging, license, or exit costs;
- waiting for production to define monitoring and support.
FAQ about cloud migration in Morocco
Should an application be rehosted or refactored?
Rehost when the priority is a fast move of a compatible workload and the remaining debt is accepted. Refactor when a verifiable business result justifies architecture change. The decision can be progressive: rehost, stabilize, then modernize a component after measuring production behavior.
Must data always stay in Morocco?
No; that blanket statement would be inaccurate. Determine the data type, sector, entity, destination, and applicable rules. Foreign transfers of personal data are governed in particular by Law 09-08 and CNDP formalities. Vital entities and infrastructures with sensitive systems or data have specific requirements under Decree 2-24-921.
How many rehearsals are needed before cutover?
There is no universal number. Rehearse until timings, reconciliation evidence, go/no-go criteria, and rollback procedures are reproducible with representative volume. A rehearsal that still reveals unassigned manual steps calls for another validation.
Is a hybrid architecture a failed migration?
No. Keeping a workload on premises can be rational because of latency, equipment, licensing, regulation, or unresolved risk. Document its flows, security, monitoring, and review date.
How should success be measured after migration?
Use the criteria defined at the start: useful availability, restore time, deployment lead time, incidents, performance, allocated cost, user feedback, and actual retirement of the old environment. A metric is useful only when its source, owner, and decision threshold are clear.
Turn migration into a controlled decision
Cloud migration in Morocco becomes defensible when every application has an outcome, strategy, test evidence, owner, and rollback path. The ten decisions create a go/no-go file that business, IT, security, finance, and leadership can review together.
Kanteek can support framing, the technical foundation, automation, and operations through its Cloud & DevOps service. The first useful deliverable is not a promise of fast migration. It is a verifiable map of workloads, risks, decisions, and expected evidence.
