Migrating to Huawei Cloud: A Practical Roadmap for Enterprises
Why a roadmap matters
Cloud migration succeeds or fails before the first workload moves. The projects that go smoothly all share the same trait: a clear, honest map of what's moving, in what order, and how each workload will be handled.
This guide is the same roadmap CloudPro uses on enterprise engagements for Huawei Cloud migrations across Egypt and MENA.
Phase 1 — Assess
Before touching anything, you need answers to four questions:
- What do we run today? A complete inventory of VMs, databases, middleware, integrations, and external dependencies.
- How is it used? CPU/RAM/IO patterns, peak vs. average load, business criticality, RPO/RTO requirements.
- What does it cost today? Real numbers — licensing, hardware refresh, data center, networking, ops headcount.
- What does it cost on Huawei Cloud? A workload-by-workload TCO model, not a sticker-price comparison.
Tools we use: Huawei Cloud MgC (Migration Center) for discovery and dependency mapping, plus our own assessment templates for cost modeling.
Phase 2 — Plan with the 6 Rs
Every workload gets classified into one of six migration strategies:
| Strategy | When to use it | Effort |
|---|---|---|
| Rehost (lift-and-shift) | Move as-is to ECS/Flexus | Low |
| Replatform | Minor tweaks (e.g., RDS instead of self-managed MySQL) | Medium |
| Refactor | Re-architect for cloud-native (containers, serverless) | High |
| Repurchase | Move to SaaS equivalent | Varies |
| Retire | Decommission — nobody uses it | Low |
| Retain | Keep on-prem (compliance, latency, hardware lock-in) | None |
A typical first wave is 70% rehost, 20% replatform, 10% refactor. You can always refactor later, on the cloud.
Phase 3 — Migrate
Migration tools we lean on heavily:
- SMS (Server Migration Service) — block-level replication of Windows/Linux VMs to ECS or Flexus.
- DRS (Data Replication Service) — online migration of MySQL, PostgreSQL, SQL Server, Oracle, MongoDB databases with minimal downtime.
- OMS (Object Storage Migration) — bulk transfer to OBS from S3, on-prem NAS, or other object stores.
- CDM (Cloud Data Migration) — for data lakes and warehouses.
The cutover playbook
- Replicate continuously to Huawei Cloud (days or weeks ahead).
- Schedule a short maintenance window.
- Stop writes to the source.
- Final delta sync.
- Cut DNS / load balancer to Huawei Cloud.
- Smoke tests.
- Keep the source running, read-only, for 7–14 days as a rollback safety net.
Phase 4 — Optimize
Post-migration is where the savings show up:
- Right-size every instance after 2–4 weeks of real production load.
- Reserve committed capacity for steady workloads (up to ~40% off).
- Autoscale elastic tiers (web, queue workers).
- Refactor incrementally — move stateless services to containers (CCE), move scheduled jobs to FunctionGraph.
- Harden security — turn on HSS, WAF, Anti-DDoS; enable Cloud Trace Service for audit logging.
A realistic timeline
For a mid-sized enterprise (50–200 servers), a typical CloudPro engagement runs:
- Weeks 1–3 — assessment & TCO model
- Weeks 4–6 — landing zone (VPCs, IAM, networking, baseline security)
- Weeks 7–14 — wave 1 migrations (low-risk workloads)
- Weeks 15–24 — wave 2 (production systems with proper change management)
- Ongoing — optimization & FinOps reviews
Next step
If you're starting to plan a migration, the most valuable thing we can do is run a free assessment workshop on your top 5 workloads. Book a free consultation and we'll set it up.
Book a free consultation with CloudPro engineers. We'll review your workloads and build a TCO model — no commitment required.