Disaster Recovery options for virtual infrastructures

Business Continuity and Disaster Recovery (BC/DR) remains a hot topic; at the end it’s all about the availability of your data and functionality. Previously I’ve blogged about solutions like VMware’s Site Recovery Manager, Zerto Virtual Replication, vSphere Replication and different storage replication technologies. I thought that it would be nice to write a little overview on what’s going on in this space. Microsoft is doing some interesting stuff as well, so let’s have a closer look. In this article I will explore VMware, Microsoft and Zerto DR solutions.

First I will explain what VMware is offering for their vSphere platform. vSphere Replication is the product that is used for hypervisor based replication. vSphere Replication the foundation for vCloud Air Disaster Recovery and also used in Site Recovery Manager (SRM) The latest version of vSphere Replication is version 6, which includes some nice improvements like network compression, support for a maximum of 2000 replications and support Linux Guest OS quiescing. You can use vSphere Replication (vSR) on its own, or use it in conjunction with SRM or vCloud Air DR.

To use vSR functionality you have to deploy special vApps that fully integrate with the vSphere WebClient. Automating a failover that uses vSphere Replication requires SRM or vCloud Air, currently there’s no public vSR public API available, although there are some rumors that an API is developed. vSR supports RPO’s of 15 minutes to 24 hours. Currently consistency groups are not a feature of vSphere Replication.

Site Recovery Manager is VMware’s solution for Disaster Recovery orchestration. SRM of course supports vSphere Replication, and/or you can choose for Array Based Replication.  A wide variety of storage replication technologies (including solutions of NetApp, EMC, HDS) are supported by SRM. The latest version of SRM is version 6, which is the evolution of SRM 5.8. Since 5.8 SRM is fully integrated with the vSphere WebClient. Currently there’s also a vRealize Orchestrator plugin available for SRM. SRM 6 fully integrates with vSphere 6 and also leverages the Platform Services Controller. Verify you’re using a supported PSC deployment scenario when implementing SRM.

Since march this year, an update to the vCloud Air Disaster Recovery service is available. vCloud Air DR is another BC/DR solution offered by VMware. vCloud Air DR now includes a fallback option, multiple point in time recoveries and some automation options leveraging a vRealize Orchestrator plugin. This last option now allows you to automate a failover to vCloud Air, similar to what SRM is doing.

As an alternative you can always choose to configure a vSphere Metro Storage Cluster (or stretched cluster); in this scenario you stretch the storage solution between the two sites. vMSC lacks the advanced orchestration options of SRM, but adds some interesting (storage) vMotion options between the sites. Note there are some limits on the distance between your sites because of latency requirements. Also read this article I’ve published last year: Future of Disaster Recovery with NextGen VMware Site Recovery Manager.

Microsoft is offering HyperV replica for the HyperV virtualization platform. HyperV replica is hypervisor based replication and built-in as a role in Windows Server 2012 R2. You can use HyperV replica (HR) for both stand-alone HyperV hosts as well as HyperV hosts which are part of a cluster and/or SCVMM. The technology support replication intervals of 30 seconds, 5 minutes or 15 minutes. You can use Hyper-V Replica to test replication without disrupting your regular production replication. There’s also an option called extended replication. In this case you can create more than one replica of the same virtual machine on different locations.

Automating a failover in a Microsoft environment involves some (custom) powershell scripting or using Microsoft Azure Site Recovery (ASR). With ASR you can automate a failover from one site to another site, the DR orchestration engine is running in the Azure cloud. This scenario both supports HyperV replica and storage based replication for HyperV. You can also choose to replicate your VMs to the Azure cloud and run failover to Azure when a disaster strikes. ASR also supports VMware host; in this case an in-guest agent is deployed. This agent can replicate data from the primary site to a secondary site, or you can replicate the data to the Azure cloud.

A third party option is Zerto Virtual Replication (ZVR). ZVR is already around for a couple years and initially only supported VMware environments. ZVR also offers hypervisor based replication, in this near synchronous so you’re RPO is almost zero. This is difference with the hypervisor based replication technologies offered by Microsoft en VMware. The latest version of ZVR is version 4.0, available since may this year.

ZVR includes hypervisor based replication but is also equipped with some DR orchestration features. You can create so called Virtual Protection Groups (VPGs); in a VPG you can define what the startup order is for the VMs in the VPG. You can choose to do a test-failover, a planned migration or an actual failover. ZVR also support consistency groups, which guarantees that a group of VMs is using the same point in time in case of a failover. Read more about Zerto in my article “A closer look at Zerto Virtual Replication 4.0“. Note that Zerto both supports Microsoft and VMware virtual infrastructures, you can even replicate and failover your virtual machines from Microsoft to VMware and vice versa.

I hope this was helpful, stay tuned for an upcoming article on VMware’s DRaaS solution vCloud Air Disaster Recovery.