Backup and redundancy continue to be two of the most common usage scenarios for public cloud -- and the reason is simple economics. Renting or building space in a secondary facility for backup is expensive, particularly when there's a thriving, competitive industry devoted to providing rentable IT infrastructure.
Due to its integration with Windows Server, and flexible licensing models for Windows workloads, Azure is a common cloud service for Microsoft-centric organizations. Below are some best practices for using Azure for business continuity (BC), disaster recovery (DR) and replication.
An overview of Azure Site Recovery
Microsoft rolled out Azure Site Recovery two years ago. The service automates the process of data and virtual application replication to back up private Windows infrastructure. But, most importantly, it provides that same application orchestration to the Azure public cloud.
Azure Site Recovery provides six primary features for backup as a service and disaster recovery as a service (DRaaS):
- Automated data protection and replication of VMs;
- Remote health monitoring of the DR installation;
- Customizable DR plans;
- Nondisruptive recovery plan testing; and
- Automated failover and recovery.
Azure Site Recovery doesn't only work with Microsoft hosts; it can replicate on-premises VMs from Hyper-V and vSphere or Windows and Linux servers to Azure or a secondary, private data center. It also supports a variety of replication methods, including application-consistent snapshots, near-synchronous replication and SQL Server AlwaysOn availability groups. Application recovery is defined by a set of scripts, Azure runbooks, or manual steps that are bundled into a package and can be automatically triggered.
Azure Site Recovery integrates with most Microsoft server applications including SharePoint, Exchange, Dynamics, SQL Server and Active Directory, and also integrates with products from Oracle, SAP, IBM and Red Hat.
BC planning and guidance
A DR/BC plan is idiosyncratic to an IT organization and its applications. Organizations should prioritize applications by business importance, including how they could potentially disrupt revenue, critical business processes, customer support and productivity.
After identifying and prioritizing applications, there are several other steps organizations should take:
- Set a recovery time objective and recovery point object for each IT workload;
- Develop data replication processes. Databases can typically use their native replication features to create block-level mirrors, while Azure Site Recovery can automate the copying of VM images. There are various techniques for copying unstructured data to Azure Files or Blobs, but the most basic approach is to schedule scripts that use Azure's AzCopy command;
- To minimize VM server sprawl, standardize system and software configurations and keep one-off permutations to a minimum; and
- Consider migrating email, messaging and collaboration users to online services, such as Office 365, Exchange Online, SharePoint Online, Google Apps or Slack, to keep communication lines open in an emergency.
DR testing in Azure Site Recovery
Azure Site Recovery provides additional automation that facilitates frequent drills without adversely affecting production environment. For example, IT organizations can run different scenarios, such as planned outages for system maintenance and unplanned disasters, to understand the data loss implications. Azure Site Recovery also automates failback to test the integrity of processes and data when restoring from Azure to the primary data center.
Azure Site Recovery caveats
Services like Azure Site Recovery simplify the cloud DR process and are great for larger organizations with cloud expertise. Smaller organizations, however, should also consider software as a service DR products, such as HotLink, Infrascale and Zerto.
Replicating all application data and system images to the cloud also creates added traffic on WAN links. During the planning process, assess existing WAN capacity and incremental replication traffic. Organizations with large amounts of data to synchronize should consider procuring a private, direct connect circuit, such as Azure ExpressRoute or AWS Direct Connect, to eliminate network bottlenecks, increase security and simplify cross-cloud network configuration.
The top Azure management tips so far in 2016
Evaluate container services from Azure, Google and AWS
Microsoft delays Azure Stack rollout