Single service comparison

Azure Backup is a single fan-in control plane spanning cloud VMs, on-prem agents (MARS/MABS), and PaaS data sources, all governed by one Azure Resource Manager vault resource. Huawei splits protection across CBR (the current unified service for ECS/EVS/SFS Turbo/desktops/files) and CSBS, the legacy server-level backup whose vaults now appear inside the CBR console; CSBS is effectively subsumed, so architects must plan against CBR and treat CSBS references as feature-parity caveats, not a separate product. Azure's vault model also unifies RBAC, policy, monitoring, and reports across heterogeneous sources, while CBR scopes vaults per resource type (server, disk, SFS Turbo).

API and data-model alignment differs materially. Azure exposes Azure Resource Manager types (vaults, protection policies, protected items, recovery points) and REST/PowerShell/CLI/Terraform coverage, plus workloads like Azure Files share backup and Blob/managed-disk backup that have no exact CBR equivalent. CBR exposes vaults, policies, and backups via its own API and console, supporting server/disk/SFS Turbo backups, application-consistent backups via custom pre/post scripts (documented for MySQL and other Linux apps), cross-region replication of backups, WORM-style backup locking, and image creation from backups. There is no first-party CBR equivalent for on-premises agent-based backup or for Azure Blob/Files native backup tiers.

Scaling, HA, and operational responsibility also diverge. Azure Backup relies on Azure-managed GRS/GZRS storage, multi-AZ Recovery Services vaults, built-in soft delete, immutable vaults, and centralized Recovery Services reports and Azure Monitor integration. CBR provides multi-AZ redundancy, cross-region replication, ransomware protection integrated with HSS, instant restore with lazy loading, and per-vault locking, but durability is realized through underlying OBS/EVS and replication features that the architect must explicitly configure. Quotas (vault count, backup frequency, retention bounds) and the replication topology are set per Huawei region, so operational runbooks, alerting, and DR drill procedures must be rebuilt rather than ported.

Assess each Azure Backup source separately: classify by workload (Azure VM, on-prem agent, SQL/SAP HANA, Files/Blob/Disk) and by protection objective (RPO/RTO, retention, cross-region/GRS, immutability). Map Azure VM and on-prem-file protection to CBR server/disk vaults; map on-prem agent workloads to a Huawei-side agent strategy (e.g., CBR file backup for cloud/on-prem directories, since there is no direct MARS equivalent). For Azure Files and managed-disk/Blob native backups, expect a gap and redesign around SFS Turbo backups, EVS backups, or OSS-based replication. Document parity for app-consistent backups, retention limits, and cross-region replication before selecting CBR-only or a composed pattern.

For data and configuration migration, there is no one-click Azure-to-Huawei backup conversion and no documented tool that imports Azure Recovery Services vaults into CBR. Re-establish protection in Huawei by creating CBR vaults of the appropriate type, defining backup policies (frequency, retention, replication), and re-protecting migrated ECS/EVS/SFS Turbo resources after the workload itself has been moved to Huawei Cloud (typically via SMS/MA/IMG or storage migration). Reimplement encryption (KMS keys), RBAC (IAM policies), tagging, and alerting against Huawei IAM, KMS, CTS, and Cloud Eye rather than translating Azure policy JSON. Recreate application-consistent backup scripts using the documented pre/post-freeze hook model.

Validate before cutover with restore drills: recover an ECS, disk, or file system to a test resource and confirm crash/app consistency, RTO, and data integrity against Huawei's instant-restore and lazy-loading behavior. Test cross-region replication by replicating a vault backup to a second region and performing an out-of-place restore. Rehearse ransomware response (HSS-triggered backup, WORM/backup locking, isolated restore) and confirm alerting on backup failures via Cloud Eye and SMN. Only after drill success should you retire the Azure Backup policies and vaults; do not assume feature parity for soft delete, immutable vaults, or centralized reports without testing equivalents.

Re-plan the cost model and quotas. Azure Backup typically bills protected-instance or GB-month storage, per-job transactions, retrieval/data-transfer, and cross-region restore egress; Huawei CBR bills per GB-month per vault type (server, disk, SFS Turbo), with OBS/EVS/SFS requests, cross-region replication traffic, and transfer added separately. Recompute TCO with peak protected capacity, request volume, retention horizon, replication topology, and Huawei region egress. Watch CBR quotas (vaults, backup frequency, retention upper bounds, replication targets per region) and open support tickets to raise them before production. Adjust runbooks, dashboards, and DR documentation to the Huawei service names and APIs.