Note
This article was originally co-authored by Reinhart Partmann and myself and was first published on comdivision.com as part of a planned three-part VMware Cloud Disaster Recovery (VCDR) series.
While the original intention was to continue with articles covering deployment, configuration, failover and failback operations, only the first article was ever completed and published.
As the original publication is no longer publicly available, and neither Reinhart nor I are associated with comdivision anymore, I have republished it here for archival and educational purposes.
VMware Cloud Disaster Recovery (VCDR) is a cloud-native Disaster Recovery as a Service (DRaaS) offering. It combines VMware Cloud on AWS with a SaaS-based recovery platform, making disaster recovery accessible without the need for maintaining a dedicated recovery datacenter.
This article is the first part of a three-part series:
- Basics and planning considerations
- Initial deployment and setup
- Recovery plans, testing, failover and failback
Customers are increasingly looking at cloud-based disaster recovery as a first step in their cloud journey. One of the biggest advantages is that there is no longer a need to keep a secondary datacenter running solely for DR purposes.
Depending on your requirements, you can choose between a cold DR setup using on-demand infrastructure or a pilot light deployment for faster recovery times.
Why Disaster Recovery Still Matters
Most organizations have documented disaster recovery procedures. Creating those procedures is one thing, but testing them regularly is arguably even more important.
After all, what is the value of a DR plan if nobody knows whether it actually works?
Recovery testing has become much easier than it was years ago. Many administrators will remember the days when a DR procedure started with retrieving tapes from an offsite location before the restore process could even begin.
In today’s software-defined world, testing can be automated and executed much more frequently.
More Than Datacenter Failures
Traditionally, disaster recovery focused on complete datacenter outages. While that is still relevant, organizations now need to prepare for a broader set of scenarios.
Examples include:
- Datacenter failures
- Hardware failures
- Ransomware attacks
- Human error
- Accidental deletion
- Natural disasters
- Security incidents
Ransomware in particular has become one of the main reasons organizations revisit their backup and recovery strategy. If your organization does not yet have a documented ransomware recovery strategy, it is worth investing time in creating one.
Typical Use Cases
VMware Cloud Disaster Recovery can be used in a number of different scenarios:
- Replacing an existing DR site
- Creating a DR site in the cloud
- Recovering from ransomware incidents
- Recovering from human error
- Protection against natural disasters
- Consolidating multiple DR locations
The ability to leverage cloud infrastructure instead of maintaining dedicated hardware can significantly reduce operational overhead and costs.
What Is VMware Cloud Disaster Recovery?
VMware has offered Disaster Recovery as a Service solutions for quite some time.
Historically this was mainly based on Site Recovery Manager (SRM) combined with VMware Cloud on AWS. Following VMware’s acquisition of Datrium, the platform evolved into what is now VMware Cloud Disaster Recovery.
Anyone who has worked with SRM before will find many familiar concepts in VCDR:
- Protection Groups
- Recovery Plans
- Orchestrated failover
- Recovery testing
In many ways the concepts remain the same, but the operational model becomes much simpler thanks to the SaaS-based approach.
VMware Cloud on AWS as Recovery Platform
VCDR uses VMware Cloud on AWS as the recovery location.
The protected environment can be:
- An on-premises VMware environment
- An existing VMware Cloud on AWS SDDC
One of the benefits of using VMware Cloud on AWS is the ability to scale capacity when required. During recovery operations, additional hosts can be added automatically if storage utilization reaches predefined thresholds.
That flexibility allows organizations to start small and scale when necessary.
Costs and Recovery Models
One of the biggest decisions is choosing between on-demand recovery and a pilot light deployment.
On-Demand (Cold DR)
With an on-demand deployment, the recovery SDDC does not exist until you need it.
Advantages:
- Lowest cost
- No permanently running recovery environment
- Only pay for the infrastructure when it is required
The trade-off is recovery time. Provisioning the recovery environment can take some time before workloads become available.
Pilot Light
With a pilot light deployment, the recovery SDDC is already available.
Advantages:
- Lower Recovery Time Objective (RTO)
- Faster failover
- Infrastructure already running
Some organizations make use of the pilot light environment for development or testing workloads during normal operations and repurpose it during a disaster event.
Core Components
Scale-Out Cloud File System (SCFS)
The Scale-Out Cloud File System acts as the cloud storage layer.
Protected virtual machines are replicated to the SCFS using encrypted, forever-incremental snapshots.
One notable advantage is that virtual machines can be recovered directly from the stored data without requiring extensive rehydration processes.
Orchestrator
The Orchestrator provides the SaaS user interface and acts as the control plane for the solution.
It is responsible for:
- Recovery orchestration
- Policy management
- Monitoring
- Automation
DRaaS Connector
The DRaaS Connector is a virtual appliance deployed within the protected environment.
Its responsibilities include:
- Snapshot replication
- Communication with cloud services
- vCenter integration
- Recovery operations
Protection Groups
Protection Groups determine:
- Which VMs are protected
- Snapshot frequency
- Retention policies
DR Plans
DR Plans define how workloads are recovered.
This includes:
- Recovery order
- Startup dependencies
- Network mappings
- Recovery workflows
How It Works
At a high level, the workflow is relatively straightforward.
- Virtual machine snapshots are replicated through the DRaaS Connector.
- Data is stored in the Scale-Out Cloud File System.
- Recovery Plans are created in the Orchestrator.
- VMware Cloud on AWS acts as the recovery destination.
- During a failover, workloads are mounted and started according to the configured recovery plan.
One useful capability is that the SCFS can be mounted into the recovery environment as an NFS datastore. This allows virtual machines to be started quickly during recovery operations.
Additional Features
A few additional features worth mentioning.
High-Frequency Snapshots
Recovery Point Objectives (RPOs) can be as low as 30 minutes, helping reduce potential data loss.
Hub-and-Spoke Deployments
Multiple protected sites can replicate into a single Scale-Out Cloud File System.
This allows multiple environments to leverage a centralized recovery platform.
Bandwidth Throttling
Replication traffic can be throttled to prevent it from impacting production workloads.
Our Thoughts on VCDR
A disaster recovery solution should be simple to understand and operate. During an incident, complexity is rarely your friend.
VCDR does a good job of abstracting much of the underlying complexity while still providing enough flexibility for enterprise environments.
There is of course a trade-off between cost and recovery objectives. On-demand recovery reduces infrastructure costs, while pilot light deployments reduce recovery times.
It will also be interesting to see how VMware continues evolving the platform as VMware Cloud services expand beyond their current footprint.
Conclusion
VMware Cloud Disaster Recovery provides a modern approach to protecting VMware workloads.
For organizations already invested in VMware technologies, the solution offers a familiar operational model while removing much of the complexity traditionally associated with disaster recovery environments.
The choice between on-demand recovery and a pilot light deployment ultimately comes down to balancing cost against recovery requirements, but it is good to have both options available.
Although the original plan was to continue this series with deployment and operational deep dives, only this first article was ever completed. Nevertheless, I hope the content still provides a useful introduction to VMware Cloud Disaster Recovery and the concepts behind it.