VMworld 2013: vVol update and VM-granular storage–NOW.

[UPDATED – 9/9/2013, 8:09am ET – voluntarily redacted something based on some friendly requests]

A topic on everyone’s mind is – what’s the latest on VM-granular storage?   vVols remain the plan on how to get there, but have pushed into future vSphere releases as people heard in today’s keynote.

EMC, BTW is committed to being first and best when it comes to vVol implementations – not only for supporting the “standard storage things” (provision, snapshot, replicate, dedupe, etc) at vVol units of management and granularity; but ALSO things that were just flat out impossible, like live vmotion of workloads over async-class workloads (which we demonstrated last year here).

To see the latest, the VNX team recorded a series of operations using the latest R&D builds of vSphere and VNX (and though that “looks” like a VNXe, it’s really not :-) which you can check out below:

You can download the high-rez version of this video here!

But, just because vVols aren’t here, it doesn’t mean we all can’t continue to innovate around “VM as the unit of management and operation”.

NetApp and EMC got it started with our vCenter plugins (still awesome!), and EMC has made Unisphere integration with vCenter from the bottom up universal (has anyone else taken this idea of embedding vCenter API integration in their native array UI and management model yet?).   Both of those idea are about VM-level management and visibility.

Everyone is working hard on VAAI on block and NAS.  Some of the NAS VAAI operations are VM-granular acceleration options.

Tintri is of course laser-focused on this use case and has a very good implementation of VM-level snaps and now remote replicas (of course they are a VMware-only array, and have their own list of what’s missing in their solution, but that’s for the market to judge, not me).

All this without vVols (again, not saying vVols “bad” – they will be an even BETTER way to do some of these things, where we are all innovating workarounds).

One interesting one – VERY INTERESTING – is what I’m about to reveal.

2 years ago, we created a virtual Recoverpoint Appliance prototype, and started work to create VM-level Continuous Data Protection using early vVol prototypes.   After all – this would be the architecturally ideal way to implement this.   Dreamy idea.   A software-only Recoverpoint – defensibly the most capable replication technology on the market:

• Rich local and Remote “Tivo” like capability – not just snaps and replicas.
• Best-in-class efficiency around optimizing replication, using dedupe, compression, and avoiding replication where possible altogether – nothing comes close.
• Broad replication topologies for fan-in/fan-out, multisite, etc.
• RPOs from async, to near-sync, to sync.  You name it.
• Consistency technology, were objects can be replicated together
• Awesome scaling – up to thousands of devices and consistency groups

This prototype would add a killer last pair of bullets:

• Completely software only solution that would work with ANY underlying storage (including vSAN!)
• … and be able to do it all with VM-level granularity.

This could be a “vSphere Replication on steroids” for another awesome example of “Software Defined Storage”.

[REDACTED – some of the innards of how we are doing this]

Virtual Recoverpoint Appliances (vRPAs) are now GA in Recoverpoint 40, which is awesome.   In addition, we are sharing the vSCSI filter now with Recoverpoint customers in beta for feedback, and demonstrating it at VMworld.   Check it out:

You can download the high-rez version of this video here!

[UPDATE: Want the bits?   Email [email protected] !!!]

Are you an EMC Recoverpoint Customer?  Do you like Recoverpoint?   What do you think of this?  Do you want to try it out?

Comments

[Travis F]

We're using RecoverPoint and very pleased with it's protection and reliability.

Regarding vSCSI, the manageability of replication granularity at the VM level and added simplicity / manageability from the VMware interface is a POWERFUL & much welcome improvement. How do we get our hands on it? Will it work with SRM?

How can we obtain vRPA's if we're currently using v3.5?

Thanks

[Yossi Saad]

Travis,
Good to hear that you use RecoverPoint and leverage its benefits. You can use the vRPA when replicating VNX, and can download the vRPA through EMC PowerLink at no additional cost. You need to be using RecoverPoint 4.0 or later (SP1 will be released in the coming days).
Note that starting 4.0 SP1, the RecoverPoint/SE configuration is customer-installable, on physical or virtual appliances.

Regarding the ESX splitter, we have just started our Beta program. The program is currently available only to VNX customers. Anyone interested to test and provide feedback is welcome to apply at: [email protected] . We apologize in advance as not all applications could be accommodated, since we have a limited number of seats available.

Yossi Saad

[Justin Mirsky]

This is great info Chad! Any word on managing Recoverpoint environments holistically from a single interface? For example, if I am replicating individual VM's and physical VM's using RecoverPoint, will it be a single management interface for both physical and vRPA's or will there be a separation?

Any chance the management interface will be getting added into Unisphere Remote? Managing RecoverPoint, XtremSF/SW, VNX and VNXe from a single console would be quite awesome :)

[Adam]

Hi Yossi
I am an EMC/VNX customer and I can't find vRPA for download on support.emc.com

[Antonio]

Will this functionality only be supported on the virtual RPA? The virtual RPA is limited to only iSCSI. Are there plans for the vRPA to support FC? Or are there plans to introduce some of this functionality with the physical RPAs?

[Chad Sakac]

@antonio - thanks for the question! The plan is to a) make the vRPA work with a new vmkernel-embedded splitter (and therefore support any underlying storage under vSphere; b) and yes, expand to include FC support over time (using NPIV). The first has a firm date, the latter is dependent on the vSphere release schedule.