Today, timed with VMworld, EMC announced a new version of their RecoverPoint data protection / disaster recovery product: RecoverPoint for Virtual Machines. This is the product that came out of what was formerly being called Project Mercury.
The new product, takes RecoverPoint further into VMware data protection than the RecoverPoint Virtual Edition did, allowing data protection granularity at the individual VM level.
I’ll walk through the pieces, including what’s different between this product and the “full version” of RecoverPoint.
What It Is
The product name pretty much covers it: this is RecoverPoint for VMware virtual machines, nothing more, nothing less. You can’t use it to protect LUNs, volumes, or datastores. Instead, you tell it which VMs to protect and it will apply the chosen protection level to all VMDKs for those VMs.
VMs can be placed into consistency groups, so policies are applied to all VMs in the group (ideal for vApps), and you can have a group of consistency groups (called a consistency group set) with the same policies.
Since RecoverPoint for Virtual Machines works in the hypervisor layer, the data protection is completely storage-agnostic. It will work with SAN or NAS from any vendor. It will work with VMware vSAN. It will work with DAS.
To work, RecoverPoint for Virtual Machines requires four components:
- The RecoverPoint vRPA (a RecoverPoint Appliance running as a VMware virtual machine)
- A VMware vCenter plugin
- An ESXi vKernel plugin (more on this in a bit)
- A plugin for the vSphere Client (only the Web Client is supported)
The key to RecoverPoint for VMware’s storage agnosticism is the vKernal Plugin. This plugin is called the Hypervisor splitter (if you’re familiar with RecoverPoint, it fulfills the same role as the old host-based splitter).
The Splitter intercepts all writes and duplicates them, sending one copy to the VMDK in was intended for, and the other to the RecoverPoint virtual appliance to apply journaling as per the data protection policies set for that VM. Because this splitting happens at the hypervisor layer — before storage is ever touched — it doesn’t matter what the storage is. If the VM lives on a datastore supported by vSphere, RecoverPoint for Virtual Machines can be used to protect it.
Ease of Use
EMC has made the product very easy to use. In vCenter, you simply right-click a VM, then select the added RecoverPoint for Virtual Machines menu. From there, you can launch the data protection wizards that will walk you through setting the protection policies for that VM, add it to consistency groups, etc.
What It Can Do
RecoverPoint for Virtual Machines supports:
- vSphere 5.1U1 and 5.5 (support for 6.0 will be available when 6.0 is GA)
- Local Continuous Data Protection (CDP)
- Remote Replication (CRR)
- Any vSphere-supported storage options
- Consistency Groups (CGs)
- Consistency Group Sets (CG Sets)
- Adding or deleting a VMDK — RecoverPoint will automagically adjust the replica to account for the change to the VM
What it Can’t Do
At this time RecoverPoint for Virtual Machines does not support:
- Simultaneous local and remote data protection — in RecoverPoint this mode is called Continuous Local and Remote (CLR) and is the key to RecoverPoint’s extreme flexibility. Support for CLR is on the roadmap, but no clear date was available to me.
- Integration with VMware vCenter Site Recover Manager (SRM). You use either SRM or RecoverPoint for Virtual Machines, but they don’t work together (although “regular” RecoverPoint integrates with SRM).
- Fan-in and fan-out. Your only option for remote protection at this time is one instance to one instance to one instance. Fan-in and fan-out (many-to-one and one-to-many) are on the roadmap. In the meantime you can work around this by running multiple instances of RecoverPoint for Virtual Machines in your cluster (up to 32 instances are supported).
- Growing a VMDK. A change in size to the VMDK would require re-protecting the VM from scratch in order to have the change applied to the replica. (Again, on the roadmap.)
- Specifying the power-up order of VMs at the remote site. This feature of SRM is not available in RecoverPoint for Virtual Machines at this time. I’m told this will likely make it into the next release.
- Automatic failover. Most folks I know prefer automated (one click) failover to automatic (it just happens for you), preferring that a human being make the decision of whether or not to failover to the DR site, but for some uptime-critical services, people want full automation.
RecoverPoint for Virtual Machines will be licensed based on the number of VMs protected, starting at a minimum of 15. Discounts will be based on volume — purchase licenses for more VMs and the per-VM cost is less.
RecoverPoint for Virtual Machines will be available in October of this year.
Try and Buy
EMC will make a fully-functional, no-charge, no time limit version of RecoverPoint available for download completely free of charge.
Go ahead and read the above sentence again. Yes, no time limit, fully-functional, no cost.
The catch? No support unless you buy licenses. Still it’s a great way to try it out before purchasing, and anybody with a home lab will now be able to have CDP for free.
Personally, I think this is a version 1.0 of what will grow to be a great product, but it’s clearly 1.0…
I love the storage agnosticism, the try and buy, the VM-level granularity, and the ease of use.
For me, this isn’t quite “there” yet until it adds:
- Support for resized VMDKs
- Full CLR option
- Specifying start-up order for VMs
That said, I’ll be looking forward to seeing how quickly these additional features get rolled out. If I had a home lab (Yes, yes, I need to fix that ASAP), I’d be eager to download this and kick the tires the moment it becomes available.