Something New! vCloud Hybrid Service — Disaster Recovery

I think the announcement of VMware's new vCHS disaster recovery service deserves some extra attention. 

One reason is that I work in the same group that built the product, so of course I want to share some love. 

But there’s more: having been in the DR and BC space for a l-o-o-o-n-g time, I think the new vCHS RaaS (recovery as a service) brings something comparatively new to the table. 

And, more broadly, this new service clearly reflects the distinctive vCHS “works the way you do” philosophy.  

All good.

I can see three use cases for vCHS - DR.  The obvious one is as a net-new solution for important applications that should be protected, but aren’t.  There’s another market out there with folks who aren’t happy with what they’ve already got for remote recovery (costs, complexity, etc.)  And there’s a third great use case around tertiary protection in addition to an existing remote recovery solution.

My Sordid History With Disaster Recovery — And Business Continuity

When I first came to EMC in 1994, the engineers were working on a very cool new product: SRDF — the Symmetrix Remote Data Facility.  We let the engineers choose the product names back then, and it shows.  It did something very simple and very powerful — it made synchronous copies of important data at a modest distance. 

The need at the time was driven by New York financial firms who wanted to “get their data out of the city” in the event of a disaster.  I quickly learned about RPO and RTO, the important differences between disaster recovery and business continuity, latency, transactionally consistent applications and all sorts of related esoterica. 

But — way back when — this was incredibly complicated and expensive stuff.   That was OK: as I quickly found out, the stakes were much, much higher for the early adopters.

Since then, the technology has become more cost-effective, and much easier to deploy.   More IT shops found they could afford to remotely replicate to a remote site to protect against all manner of natural and man-made disasters.  But — even to this day — setting up and testing a remote recovery capability can still be a complicated and expensive proposition.

For one thing, most of the technology solutions are essentially bolted-on — a separate set of products and processes are required to completely protect and recover even a modest application landscape.  And then -- if you didn't want to build your own data center -- there were the daunting economics associated with finding a hosting provider or colo, paying for capacity you hoped you’d never have to use — except for testing.  

The net result is that there’s a vast ocean of under-protected applications out there that — by all rights — could easily justify remote protection, especially if it was very simple and cost-effective.

That’s what I really like about this new vCHS - DR offering.  It’s about as dead-simple as possible to set up and test.  The economics look compelling.  And - best of all - it’s fully integrated into the vSphere environment from an operational perspective - it works the way you do, so there’s not much to learn.

I think it’s going to end up being very popular.

The Basics

The new vCHS-DR uses vSphere Replication as its basic engine to make remote copies.  If you’re not familiar, VR is an async delta-based replicator that’s fully integrated into vSphere and its management tools. 

The RPOs (recovery point objectives) are not particularly aggressive — in the range of 15 minutes to 24 hours — but certainly sufficient for the intended use cases.  RTO is committed to be less than four hours — again, more than sufficient.

On the back end, VMware has created a new kind of DR instance as a starting point, which can be extended as needed: more resources, more IP addresses, etc.  The press release mentioned an $835/month starting price, which includes a basic setup, two tests per year, reserved capacity, as well as 30 days of remote run time as needed.  To some, that might sound pricey, but — when you actually go shopping for realistic options, it turns out to be quite reasonable.

The local side looks “built-in”.  You install a few software packages, you tell it about your vCHS account, click “next” a few times — and the protection starts immediately.  Monitoring is done in context; and there’s good automation for periodic testing.  Fail-back (should it ever be needed) is semi-automated.

Because vSphere Replication runs above the storage layer, it has some interesting attributes when compared to other alternatives.  

First, it’s completely agnostic to your choices of servers and storage.  That’s important, as so many existing DR solutions insist on a like-for-like on both ends.   This hardware independence keeps costs down, and delivers much more flexibility on the source side.

Yes, it works with VSAN :)

Second, it fully encapsulates your applications, just as vSphere does.  One of the challenges with traditional DR is making sure all the various bits and pieces make it to the remote side on a consistent basis.   As long as it runs on VMFS, it’s replicated in its entirety.

Finally, it can be used transparently on top of an existing remote replication setup as a third copy if additional protection is desired.  Many shops would like a remote bunker architecture: a short hop to a location that’s reasonably close, combined with a longer, asynchronous hop to a distant location.   The vCHS DR solution solves that second-hop requirement quite simply and elegantly.

What This Means

I think the biggest impact will be in smaller IT environments that really would like some form of remote protection, but don’t have the staff or budget.  With vCHS DR, the costs are quite reasonable, and there’s no big project or new skills requirements.  It’s pretty close to just enabling a new feature on vSphere.   

I can’t tell you how many people I’ve met over the years that wish they could better protect their critical applications, but just couldn’t muster the resources.   Maybe they can sleep better now.

Even larger shops with existing DR capabilities will also be interested, especially if they’ve been thinking about a second replication hop to a distant data center.  So many categories of disasters can be regional ones: weather, floods, power grids — even political unrest, depending on where you’re doing business in the world.

My congratulations to the vCHS team for delivering a targeted service that solves a real problem for so many people.

--------------------

Like this post?  Why not subscribe via email?

Comments

[Michael Leung]

I think I need to promote that to my clients. Some of my clients asked my company to build a DR for them for a while

[A Datta]

I have been following DRaaS market for the past one month, after reading your blog a question cropped in my mind, it wouldbe a lot of help if you can answer it.
Can you please let me know how VMware addresses the issue of multiple enterprise platforms with hybrid deployments?

[Chuck Hollis]

Hello Asmita:

I'm not clear what you mean by "multiple enterprise platforms with hybrid deployments". Could you explain a bit more?

Thanks

[A Datta]

Hello Chuck,

In your blog you talked about vCloud Hybrid being server and hardware agnostic. Does this imply that it can address multiple applications running ONLY on VMware virtualized platforms or it can support VMware+non-VMware platforms simultaneously?

Thank You

[Chuck Hollis]

The idea behind vCHS is to provide a compatible public cloud for vSphere users: same technology, same management tools, same operational experience, etc.

So, no, it would be vSphere only.