RecoverPoint as a Virtual Appliance (vRPA)

« Charlotte VMUG will be a doozy | Main | EMC Atmos Virtual Appliance both to play, and for production »

June 02, 2010

RecoverPoint as a Virtual Appliance (vRPA)

One killer breakout at EMC World was “HOWTO configure a Virtual RecoverPoint Appliance.

This takes one of the most sophisticated replication (both local and remote) products on the market and makes it simple to install and configure as a virtual appliance for non-production uses.

Making something a product (with everything it entails) supported in production environments takes more time (before you ask, YES we are working on it :-), but this makes it easy for people to play, try, learn, demo.

UPDATE (missed an important person): Big thanks to Ken Gould and Steve Hagerty from the Virtualization Solutions Center in Cork (Fellas you rock – you can always tell the Cork crew by their cool Irish brogue accent in the videos:-) and the always fantastic Stephen Spellicy on the vSpecialist squad. Extra thanks to Rick Walsworth and the Israeli RecoverPoint team – for making a kick-a$$ product (customers love it) and taking a bold step of sharing this goodness.

So, what is EMC RecoverPoint?

EMC Recoverpoint does simple, efficient (while sophisticated) local and remote continuous replication. Think of this as a “Tivo” for general purpose storage use cases. It’s not snapshots, it’s a continuous journal that can be used for local purposes or on the remote side.

Above and beyond that core function, it:

compresses the WAN traffic in real-time – customers often find 7x more efficient use of the WAN compared with traditional async array based replication. That saves big $$.
Supports a broad range of RPOs – from strict sync, async (you can select the RPO), or adaptive.

Has a scale out architecture – you can add more RecoverPoint appliances (up to 8 on each side) to increase device counts and also increase replication bandwidth. Can scale out to replicating 600TB of storage (if you’re even bigger, EMC has solutions that can go even higher).
has very sophisticated consistency group technology – can in fact have consistency groups spanning RecoverPoint appliances in a cluster.
Is deeply integrated with the vCenter APIs – you can see the relationships of VMs to replication state directly in RecoverPoint. Recent additions in v3.3 make this even better – with more comprehensive logging and filtering for VMware-related tasks. See the screenshot below where you can see whether individual VMs are replicated or not, and their relationship with VMware datastore objects. If only I could share where we were going next re: RecoverPoint/VMware integration – it’s super-cool :-)

Great SRM support – recent additions in v3.3 make this even better – supporting use cases where the failed over state needs to be maintained over a long period with loads of data change. RecoverPoint also has an automated SRM failback capability like EMC’s other replication capabilities.
Can support heterogenous configurations (non-EMC arrays)
In Unisphere (the unified next-generation EMC GUI), RecoverPoint is integrated in with primary file and block storage management UI as shown here.

If you’re interested in trying out RecoverPoint, or are an EMC partner (technology partner or reseller), the vRPA makes this as simple as can be.

If you’re interested, read on!

Ok, first of all, here’s a step-by-step HOWTO guide on getting it setup.

To use this, you will need a RecoverPoint ISO. This should be available to EMCers, customers and EMC partners on Powerlink here - Home > Support > Software Downloads and Licensing > Downloads P-R > RecoverPoint and RecoverPoint/SE

Note that there are also minor hardware pre-requisites. You need Nehalem-based CPUs, vSphere 4 (both needed for VMdirectpath) and Qlogic 24xx/25xx HBAs.

Editorial note: Normally, as people who know me can attest to, I’m not a general fan of VMdirectpath. Most people think of it as a “performance thing” when the reality is that with VMXNET3 and pvSCSI adapters in vSphere 4 you can get to within low percentage points of the same performance as you can with “hypervisor bypass” (met 3 customers in the last 2 weeks who were adamant they needed VMDirectPath for “performance” – but didn’t know what their perf goals were, and were using VI3 as the baseline - argh), and I know that this is called out in Cisco’s training on UCS. For some reason people think for “high performance IO” you can’t do it without bypass, but that is not explicitly true. Also people conflate interface virtualization (SR-IOV) with VMdirectpath. One doesn’t necessitate the other.

This use case for VMdirectpath is interesting and has nothing to do with performance. It’s needed because the RPA explicitly expects those qlogic adapters, and VMdirectpath exposes those up transparently to the vRPA. VMDirectPath made virtualizing an appliance with a hardware dependency much easier.

Something to understand when you’re reading the doc. RecoverPoint creates a copy of every IO as it travels from the host to the target and stores this in a journal. Where this “splitter” resides can vary choices are:

In the Array itself. This is embedded in EMC’s midrange Unified platform and is the simplest (and highest customer satisfaction) way of doing it. By definition, this homogenous configurations only.
In the fabric. Both Cisco SANTAP and Brocade Intelligent Fabric support splitting off Recoverpoint IO. This supports heterogeneous configurations. It is notable that these configurations are much more complex (upside is they are much more flexible), and you need to follow the interop guideline very closely.
On the host. Recoverpoint has a host-splitter, which is a filter driver, with support for a broad set of OSes.

Ok – not much of a reader? Here’s the video series highlighting how easy this really is.

HOWTO: Installing a Virtual RecoverPoint Cluster - Part I: covers the basics of getting vRPA installed.