It is vital to investigate the disaster recovery capabilities of any cloud provider to evaluate its DR scheme against...
your mission-critical workloads' requirements. The last thing you need is for your most crucial workloads to crash. Every public cloud vendor will have different protocols. As an example, Microsoft Azure incorporates redundant virtual machine instances across multiple fault domains -- which are groups of racks or different Azure data centers -- to ensure that a fault on one Azure server will not completely disable a critical workload.
However, just having multiple copies of a workload does not guarantee satisfactory behavior during failover or recovery activities. Part of the evaluation for any cloud provider should include a more comprehensive assessment of its disaster recovery capabilities -- including any impact of workload replication to remote data centers. And remember that replicating some critical workloads might violate geolocation compliance rules, so know where your important workloads live and be sure to get geographical commitments in writing in your service-level agreement.
In addition, some workloads might not be properly architected for availability, and this might affect performance when a system failure at the provider's end causes a failover to a remote server. This failover occurs most often in legacy or custom applications not natively designed for cloud deployments. If a provider can't guarantee DR performance, it may be necessary to forgo the public cloud and keep mission-critical workloads in-house.
And don't expect cloud providers to handle your data protection or backups automatically -- a cloud instance is not necessarily a protected instance. Backup services, when available, may be provided as an optional service, which will carry an additional cost. For example, VMware vCloud Air provides a data protection option with 1 terabyte of storage for 43 cents per gigabyte per month. Know how to access and restore protected data content.
About the author:
Stephen J. Bigelow is the senior technology editor of the Data Center and Virtualization Media Group. He can be reached at firstname.lastname@example.org.
DRaaS the key to sleeping soundly
Disaster recovery is the first step to public cloud
VMware catching up to IBM in cloud DRaaS