Business continuity and disaster recovery – Cloud Adoption Framework

Table of Contents0.1 In this article1 Design considerations2 Design recommendations Article 01/06/2022 2 minutes to read 4 contributors In this article Organization and enterprise application workloads have recovery time objective (RTO) and recovery point objective (RPO) requirements. Effective […]

Organization and enterprise application workloads have recovery time objective (RTO) and recovery point objective (RPO) requirements. Effective business continuity and disaster recovery (BCDR) design provides platform-level capabilities that meet these requirements. To design BCDR capabilities, capture platform disaster recovery (DR) requirements.

Design considerations

Consider the following factors when designing BCDR for application workloads:

  • Application and data availability requirements:

    • RTO and RPO requirements for each workload.
    • Support for active-active and active-passive availability patterns.
  • BCDR as a service for platform-as-a-service (PaaS) services:

    • Native DR and high-availability (HA) feature support.
    • Geo-replication and DR capabilities for PaaS services.
  • Support for multiregion deployments for failover, with component proximity for performance.

  • Application operations with reduced functionality or degraded performance during an outage.

  • Workload suitability for Availability Zones or availability sets:

    • Data sharing and dependencies between zones.
    • Availability Zones compared to availability sets impact on update domains.
    • Percentage of workloads that can be under maintenance simultaneously.
    • Availability Zones support for specific virtual machine (VM) stock-keeping units (SKUs). For example, Azure Ultra Disk Storage requires using Availability Zones.
  • Consistent backups for applications and data:

    • VM snapshots.
    • Azure Backup Recovery Services vaults.
    • Subscription limits restricting the number of Recovery Services vaults and the size of each vault.
  • Network connectivity if a failover occurs:

    • Bandwidth capacity planning for Azure ExpressRoute.
    • Traffic routing during a regional, zonal, or network outage.
  • Planned and unplanned failovers:

    • IP address consistency requirements, and the potential need to maintain IP addresses after failover and failback.
    • Maintaining engineering DevOps capabilities.
    • Azure Key Vault DR for application keys, certificates, and secrets.

Design recommendations

The following design practices support BCDR for application workloads:

  • Employ Azure Site Recovery for Azure-to-Azure VM DR scenarios.

    Site Recovery uses real-time replication and recovery automation to replicate workloads across regions. Built-in platform capabilities for VM workloads meet low RPO and RTO requirements. You can use Site Recovery to run recovery drills without affecting production workloads. You can also use Azure Policy to enable replication and to audit VM protection.

  • Use native PaaS DR capabilities.

    Built-in PaaS features simplify both design and deployment automation for replication and failover in workload architectures. Organizations that define service standards can also audit and enforce the service configuration through Azure Policy.

  • Use Azure-native backup capabilities.

    Azure Backup and PaaS-native backup features remove the need for third-party backup software and infrastructure. As with other native features, you can set, audit, and enforce backup configurations with Azure Policy to ensure compliance with organization requirements.

  • Use multiple regions and peering locations for ExpressRoute connectivity.

    A redundant hybrid network architecture can help ensure uninterrupted cross-premises connectivity if an outage affects an Azure region or peering provider location.

  • Avoid using overlapping IP address ranges for production and DR sites.

    Production DR networks that use the same classless interdomain routing blocks require a failover process that can complicate and delay application failover. When possible, plan for a BCDR network architecture that provides concurrent connectivity to all sites.

Source Article

Next Post

MLOps: ML model management v1 - Azure Machine Learning

Mon Aug 1 , 2022
Table of Contents0.1 In this article1 What is MLOps?2 MLOps in Azure Machine Learning3 Create reproducible ML pipelines4 Create reusable software environments5 Register, package, and deploy models from anywhere5.1 Register and track ML models5.2 Package and debug models5.3 Convert and optimize models5.4 Use models5.4.1 Batch scoring5.4.2 Real-time web services5.4.3 Controlled […]
Exit mobile version