By: Ashley Neely
Yesterday, we announced the new vCloud Air Dedicated Disaster Recovery solution, a set of vCloud Air services designed to provide enterprise customers a highly scalable and flexible DRaaS solution that can replace the need for a secondary data center. Part of what makes this solution powerful is that customers have the ability to protect their on-premises vSphere environment in multiple ways depending on a workload’s required Recovery Point Objective (RPO) and Recovery Time Objective (RTO) targets. The other key enhancement is the improved performance and scalability that this solution can achieve by incorporating Hybrid Cloud Manager and our Hybrid DMZ reference design. In this blog we will explore the multiple protection techniques that are supported with this service.
Protecting Your Data Center in the Cloud
Most vSphere customers have a broad set of applications in their on-premises data center from mission critical workloads to test and development workloads. When it comes to protecting those workloads and ensuring business continuity through a major event, different workloads will have different needs. With vCloud Air Dedicated Disaster Recovery solutions, we can offer three primary methods of protecting your data center workloads:
- Leverage vSphere Replication over a WAN-optimized connection for improved performance
- Disaster avoidance through planned migrations, etiher with zero or low downtime
- Support active-active application availability where available
By leveraging the SD-WAN capabilities of Hybrid Cloud Manager, this new solution is able to improve upon the standard vSphere Replication application appliance. Below is a basic architecture which connects your data center directly to a vCloud Air Dedicated Cloud instance. It does not incorporate a Hybrid DMZ, but we will be going into some of these other commonly used architectures in a later post.
As this diagram highlights, Hybrid Cloud Manager forms a secure connection between your data center and vCloud Air. This WAN-optimized connection allows replication traffic to flow much faster to your vCloud Air instance, allowing you to achieve improved RPOs.
Replication-based DR provides two different types of protection. As data is constantly being synchronized with a copy that is sitting in vCloud Air, if your primary data center goes down, you will always have a copy to recover from in the cloud. If your RPO target was 15 minutes, the copy in vCloud Air will be at most 15 minutes out of sync with what was running live. Sometimes, however, you will be able to execute a planned cutover – especially if you have warning or time to initiate the planned migration. When a virtual machine is already configured for replication and needs to be immediately moved to the cloud side, choosing to do a planned migration will do an optional final sync and cutover.
Migrate Workloads for Disaster Avoidance
Besides replication-based DR, this new solution provides new methods of disaster avoidance by taking advantage of the three migration methods offered by Hybrid Cloud Manager (HCM). With my own migration of a 20GB test VM, migrations happened in less than 5 minutes. An independent review done by Principled Technologies showed up to a 70% decrease in transfer times with HCM’s WAN Optimization.
Cross Cloud vMotions (Zero Downtime Migrations)
Stretching a Layer 2 Network with HCM into your vCloud Air environment also allows for cross cloud vMotions. Like an on-premises vMotion, the VM is moved, and during this time it may drop a few pings, at most. End users should not notice any downtime. In addition to a stretched network, cross cloud vMotions also require hardware Version 9 or higher. Onecross cloud vMotion can be performed at a time, with others queued up to immediately start once the current cross cloud vMotion is completed. Once a cross cloud vMotion is completed, the VM is removed from on-premises.
Cold migrations are performed on a VM that is turned off. When the migration is completed, the VM is moved to a “Migrated VMs” folder. Cold migrations have the same requirements: HW version 9 and a stretched network.
Low Downtime Migrations
Low downtime migrations take advantage of replication technology. When configuring a VM for a low downtime migration, the VM does an initial sync then performs a delta sync before the VM is migrated. This gives us the option to set a change window to perform this last sync and cut over. In the HCM “VM Migrations“ tab, you will be able to select a quick switchover or a scheduled switchover.
Finally, there are certain applications like SharePoint and SQL Server that support Active-Active Availability groups which is an in-application technology that supports distributing workloads across different data centers for distributed protection. With Dedicated Disaster Recovery solutions, we can support mixed modes of active workloads alongside replicated and migrated workloads. That means that these workloads can co-exist directly with replicated VMs in the same Virtual Data Center.
While traditional recoveries in the cloud during an outage or disaster can easily be done, our myriad of migration options allow you to customize your recovery solution to your needs. For example, if you have 3 hours to get everything running in vCloud Air, you can start staggering scheduled low-downtime migrations for less critical VMs while simultaneously using planned migrations on already replicated VMs. Between vSphere Replication and HCM, you will be able to choose the best migration option for any scenario you are facing.