EMC World 2015: Event Recap

After EMC World 2015, I’m languishing in airports today in post-conference burnout – an ideal time to deliver a report on the news, announcements and my prognostications on what this means to our business.

The big announcements were delivered in General Sessions on Monday (EMC Information Infrastructure & VCE) and on Tuesday (Federation: VMware & Pivotal). The Federation announcements are more developer and futures oriented, although important strategically, so I’ll pass on that for now.

EMC and VCE have updated their converged and Hyperconverged products pretty dramatically. Yes, VSPEX Blue is Hyperconverged, however unfortunate the name is in linking an EVO:RAIL solution to a reference architecture solution.

The products can be aligned as:

  1. Block
  2. Rack
  3.  Appliances
EMC World 2015

The VCE Vblock product line adheres to its core value proposition closely.

  1. Time from order to completely deployed on the data center floor in 45 days. (GreenPages will provide the Deploy & Implementation services. We have three D&I engineers on staff now.)
  2. Cross component Unified upgrade through a Release Candidate Matrix – every single bit of hardware is tested in major and minor upgrades to insure compatibility: storage, switch, blade, add-ons (RecoverPoint, Avamar, VPLEX).
  3. Unified support – one call to VCE, not to all the vendors in the build

However, VCE is adding options and variety to make the product less monolithic.

  1. VXblock – this is the XtremIO version, intended for large VDI or mission critical transactional deployments (trading, insurance, national healthcare claims processing). The Beast is a Vblock of eight 40 TB Xbrick nodes, 320 TB before dedupe and compression, or nearly 2 PB with realistic data reduction. Yes, that is Two Petabytes of All Flash Array. Remote replication is now totally supported with RecoverPoint.
  2. VXRack – this is a Vblock without an array, but it isn’t VSAN either. It is….ScaleIO, a software storage solution that pools server storage into a shared pool. The minimum configuration is 100 compute nodes, which can be dense performance (4 node form factor in 2 U chassis) or capacity. The nodes can be bare metal or hypervisor of any sort. This can scale to 328 Petabytes. Yes, Petabytes. This is web-scale, but they call it “Rack Scale” computing (first generation). More on that later…
  3. Vscale – Networking! This is Leaf and Spine networking in a rack to tie a VXrack or Vblock deployment together, at scale. “One Ring to Rule Them All”. This is big, literally. Imagine ordering a petabyte installation of VXblock, VXrack and Vscale, and rolling it onto the floor in less than two months.

So, that is Block and Rack. What about Appliance?

Enter VSPEX Blue, the EMC implementation of EVO:RAIL. This has definite value in…

  • Pricing
  • Unified management & support
  • The “app store” with
    • integrated backup (VDPA)
    • replication (vRPA)
    • Cloud Array integration (TwinStrata lives!), a virtual iSCSI controller that will present cloud storage to the system as a backup target or a capacity tier.

This post from Mike Colson provides a good explanation.

Future apps will include virus scanning, links to Public IaaS and others.

I set one up in the lab in 15 minutes, as advertised, although I had to wait for the configuration wizard to churn away after I initialized it and input all the networking. Professional Services will be required, as EMC is requiring PS to implement. Our team is and will be prepared to deploy this. We can discuss how this compares to other Hyperconverged appliances. Contact us for more information.

There are other announcements, some in sheer scale and some in desirable new features.

Data Domain Beast: DD9500, 58.7 TB/hr. and 1.7 PB of capacity. This is rated at 1.5x the performance and 4x the scalability of the nearest competitor.

VPLEX News: The VPLEX Witness can now be deployed in the public Cloud (naturally EMC recommends the EMC Hybrid Cloud or vCloud Air). The Witness has to be outside the fault domains of any protected site, so where better than the Cloud? It is a very lightweight VM.

CloudArray (TwinStrata’s Cloud Array Controller) is integrated with VPLEX. You can have a distributed volume spanning on premise and cloud storage. I’m still trying to grasp the significance of this. The local cache for the CloudArray controller can be very fast, so this isn’t limited to low latency applications. The things you could do…

VPLEX is now available in a Virtual Edition (VPLEX/VE). This will obviously come with some caveats and restrictions, but this also is a fantastic new option for smaller organizations looking for the high availability that VPLEX provides, as well as data mobility and federation of workloads across metro distances.

VVOL: Chuck Hollis (@chuckhollis) led an entertaining and informative ‘Birds of a Feather’ session for VVOLs. Takeaway – this is NOT commonly deployed yet. Only a handful of people have even set it up, and mostly for test. This was in a room with at least 150 people, so high interest, but low deployment. Everyone sees the potential and is looking forward to real world policy based deployments on industry standard storage. This is an emerging technology that will be watched closely.

VNX/VNXe: I didn’t see or hear many striking features or upgrades in this product line, but an all flash VNXe was trumpeted. I’ll be looking at the performance and design specifications of this more closely to see how it might fit targeted use cases or general purpose storage for SMB and commercial level customers. There is talk around the virtualization of the VNX array, as well as Isilon, so pretty soon nearly every controller or device in the EMC portfolio will be available as a virtual appliance. This leads me to…

ViPR Controller and ViPR SRM: Software Defined Storage

ViPR Controller is definitely a real product with real usefulness. This is the automation and provisioning tool for a wide variety of infrastructure elements, allowing for creation of virtual arrays with policy based provisioning, leveraging every data service imaginable: dedupe, replication, snapshots, file services, block services and so on.

ViPR SRM is the capacity reporting and monitoring tool that provides the management of capacity that is needed in an SDS environment. This is a much improved product with a very nice GUI and more intuitive approach to counters and metrics.

I’d recommend a Storage Transformation Workshop for people interested in exploring how SDS can change the way (and cost) of how you manage your information infrastructure.

More on EVO:RAIL/VSPEX Blue

I met with Mike McDonough, the mastermind behind EVO:RAIL. He is indeed a mastermind. The story of the rise of EVO:RAIL as a separate business unit is interesting enough (300 business cases submitted, 3 approved, and he won’t say what the other mystery products are), but the implementation and strategy and vision are what matter to us. The big factor here was boiling down the support cases to come up with the 370 most common reasons for support, all around configuration, management and hardware. The first version of EVO:RAIL addressed 240 of those issues. Think of this as having a safety rail around a vSphere appliance to prevent these common and easily avoidable issues, without restricting the flexibility too much. The next version will incorporate NSX, most likely. Security and inspection are the emphases for the next iteration. Partners and distributors were chosen carefully. GreenPages is one of only 9 national partners chosen for this, based on our long history as a strategic partner and our thought leadership! The tightly controlled hardware compatibility list is a strength, as future regression tests for software and other upgrades will keep the permutations down to a minimum. (By the way, the EMC server platform is Intel, for VxRack, VSPEX Blue and I think for all of their compute modules for all their products). The implication here, competitively, is that as competitive appliances that are buying white box hardware with commodity contracts allowing for flexibility in drives, memory and CPU, will have an exponentially more difficult task in maintain the increasing permutations of hardware versions over time.

Final Blue Sky note:

Rack Scale is an Intel initiative that promises an interesting future for increased awareness of the hardware for hypervisors, but is a very future leaning project. Read Scott Lowe’s thoughts on this.

 

As always, contact us for more details and in-depth conversations about how we can help you build the data center of the future, today.

 

By Randy Weis, Practice Manager, Information Infrastructure