The best athletes (and VxBlock 1000) = faster, higher, stronger!

In watching the Olympics, it’s amazing to see athletes doing amazing things – frankly it’s inspiring.

Sometimes it’s a new star rising – something new (amazing Chloe Kim!) .  

Sometimes it’s a veteran pulling a “Michael Phelps” – producing every 4 years (see Dutch speed skater Sven Kramer).

I’m not claiming that what we do at work compares to those amazing athletes do… But it makes for a timely analogy  with the VxBlock 1000 Ready To Ship (RTS) milestone today 🙂   If you want to learn more, join the #CrowdChat at 11:55am ET today here.

While I no longer lead the efforts behind CI/HCI at Dell EMC – I still have a passion for it, and love the people and the teams in their new homes (post coming up on what’s next for me soon enough).

In 2009, Vblock was born. Cisco together with EMC invented the Converged Infrastructure category – and have led it ever since.   Dell EMC VxBlock is the “Sven Kramer” of Converged Infrastructure 🙂

The founding idea was basic, simple:  the complexity associated with the base infrastructure stack was a waste of time and money. That waste is bad enough when all you do is look at is acquisition and initial costs – but the waste balloons out when you look at lifecycle management of base infra over years – along with multi-vendor support and commercial costs.   What if we were to design, build, manufacture, support, sustain over it’s lifecycle the entire compute/storage/network domain – for bare metal, and up to the VMware layer – as ONE THING?

This complexity of siloes of infrastructure is purely sand in the gears – zero value.   It may be a familiar model to people in infrastructure siloes, or purchasing teams accustomed to optimizing for piece parts and using price arbitrage levers…  But in every SINGLE case, there’s someone who is looking at the system-level operating complexity and cost and asking “THERE MUST BE A BETTER WAY”.

There is – CI and HCI is that better way.  They are about changing what you construct into what you consume.

They say that imitation is the sincerest form of flattery.   I understand this personally.   We certainly didn’t invent HCI – but when we entered the market, we did it in force, we learnt from the people who went first, refined, leveraged our strengths and used that to start innovating on our own.   In short order, Dell EMC and VMware became #1 in HCI.   I am thankful for those who came before, and respect our competitors.  We will furiously defend our #1 position and work to broaden the position.

In CI there are many imitations of the VxBlock business – and likewise respected competitors.  Unlike HCI where a year ago we were NOT #1, we were NOT the first – and needed to fight to get the #1 spot; in CI it’s a different story.

When it comes to CI – we have been successfully defending the #1 spot – for almost a DECADE.

Many people (including some in Dell EMC and our channel partners) still view CI as a “bundle”.  Many view the value of CI as simply being the sum of the parts (UCS, NEXUS, storage partner).   We don’t – VxBlock is about changing the HOW of your operational model, not the WHAT.  

None of the CI competitors have furiously embraced the “lifecycle management” idea, and most involve “meet in the channel” business models of servers, network, storage.  

These bundles and “meet in the channel” models are different from the “single product, single lifecycle, everything from one place” foundation of VCE and the Vblock.  When we execute well on the simplicity of initial standup acceleration, and more importantly, improved lifecycle vs the “piece parts” – that’s when our CI customers are the most happy.

The fact that others have taken a different path is perfectly OK – ultimately customers decide and vote with their dollars on what matters most to them.

Now – what’s new?   In early 2017, we realized that it was time for a simplification, a re-invention of Converged Infrastructure.   The VxBlock 1000 is a huge milestone – not the first, and not the last.

The work started with going from 2 product families (Vblock and VxBlock) to one (VxBlock). This simplified the Release Certification Matrix (RCM) enormously – and use of the VMware Distributed Switch enables SDN models on VxBlock – we started to see a much, much higher prevalence of NSX on VxBlock as an example. But that was simply the first step. Today’s step is much bigger.

Today the team launched the VxBlock 1000, the next step in simplification and ultimately (over time) full re-invention.  Vxblock 1000 =  industry’s best CI just got faster, higher, stronger.  

So… what’s new?



    • With VxBlock 1000 – we go from having 3 types of VxBlock (VxBlock 350, 540, 740), three corresponding RCMs, and islands of resources and management to a single RCM.   Our existing customers (of course) can expect the existing investments to be supported for their full lifecycle…. But going forward, it’s one, simplified CI offer.
    • We’ve developed an interactive online lifecycle management portal – this can make checking and resolving low level firmware compliance take 80% less time.   People that run infrastructure know that low level stuff can really bite you.   VxBlock cuts down the risk a lot – and ultimately, we own the outcome.
    • We’ve developed a much simpler and scalable management appliance – the next generation AMP = AMP-VX. The new AMP-VX is smaller, less costly, more flexible.  AMP-VX uses vSAN and an HCI model for simplicity, and the ability to scale, and has a built in Data Domain DD2200 for localized backup of system info.   The new AMP-VX can handle up to 8 VxBlock 1000 systems, which themselves can be massive. The AMP-VX also has our next-generation remote support systems, and embedded data protection.   You heard that right – beyond the data protection of the AMP-VX, every VxBlock 1000 include a trial of the Converged Infrastructure Data Protection suite, which can be turned on for the whole system and all workloads if you love it.
  • You have 4x the choice in the storage layer (XtremIO X2, VMAX AF, Unity AF, Isilon Gen 6)
  • You have 2x the choice in the compute layer (UCS b-series M4 and M5, and c-series) – meaning you can pool VDI dense systems with GPUs, along with test/dev and prod VMs, containers and other very different workloads.
  • … All with the value of system-level acquisition, lifecycle and support.  You don’t need different “islands”, or Vscale (except at the most massive datacenter scales)
    • CI done right isn’t about the ingredients – except for insofar that those ingredients change the “shape” of the system’s economics, performance or other parameters.
      • The VxBlock 1000 can be configured with a broad range of the following:
        • the new Unity X50F – which has a 4x greater density/performance vs. the previous generation
        • XtremIO X1 to X2 = 2 to 1 footprint reduction
        • Isilon Gen 5 to Gen 6 = 6 to 1 footprint reduction… And Isilon Gen 6 can be built-in, no need for a tech extension.
        • the new M5 UCS B-series blades (a VxBlock 1000 can support up to 800 blades up from the previous maximum of 512), and upgraded 40G network with 100G uplinks, and more.
    • We’ve designed the VxBlock 1000 to be NVMe ready – as we know that customers will be deploying VxBlock 1000 and will have them up and running over a service life that may very well include NVMe. The fabric is ready to integrated NVMe storage initiators and targets when you need to go from milliseconds to microsecond timescales.
    • The smarts in the AMP-VX are just the start.   VxBlock 1000 fits into Vscale, can integrate with VxRack systems, including fabric automation.   While there’s nothing stopping customers from using the element managers that they know – we know that system-level automation is an area of continued opportunity for innovation.

A couple real examples from customer scenarioes can help make this leap forward more visceral.  

Note that these may strike you all as BIG.   They are BIG.   I’ve often joked that VxBlock is available in any size you want – so long as it’s large, XL, or XXL.   CI fundamentally doesn’t scale DOWN well.   The economics of CI are best at moderate scale and up.   You also cannot physically make a CI system that is physically really small… This is one of the HCI advantages.

The first example is a very common model – standard general purpose VSI (virtual server infrastructure).   At the top is a VxBlock 350 example – and our CI competitors would be similar.    VxBlock 1000 has a 25% smaller footprint, and similar reductions in capital and operation expenses.   Why?  


Simple.   CI normally scales to a certain point, and then you get another “block”.   Anyone who says “you can scale it however you want” is smoking something, or is not on the hot seat for sustained lifecycle and support.   Why do I say that?  There are system design parameters (like fabric capacity, buffer credits, initiator/target maximums, and many, many others) that go into a system level design.      While VxBlock 1000 has maximums (it must of course, it’s maximums are so much higher) the multiple systems can be collapsed.

Here’s another example.   It’s not uncommon for customers to have multiple “tiers” of VxBlocks based on RAS (Reliability, Availability, Serviceability) and specific data services.   This IS NOT for general purporse VM workloads – but workloads with very specific requirements.   If you are nodding – you are a customer like this.  If you are looking at this post quizzically, you are not one of these customers.   These customers often have Multiple VxBlock 740s and several VxBlock 350s, and a VxBlock 540 or two.   This kind of customer ALWAYS has a need for a large scale-out NAS use case – and want it in the same single lifecycle, “single throat to choke” model.


In this case – there’s a massive improvement.   First, no “islands” of compute or storage capacity as workloads move from test/dev into prod.   There’s a shared fabric for everything (resulting in a massive 70% reduction in networking cost).   ALL the systems fit into one system, and one management pod (one AMP-VX).   When they need to do a lifecycle update – it’s a 75% improvement.      Try to imagine doing this yourself, with a mishmash of kit – or even with other forms of CI.

VxBlock 1000 is ORDERABLE TODAY. 

All of these improvements are built on top of time-tested, customer validated VxBlock value.   VxBlock 1000 builds upon documented VxBlock benefits (from “The Business Value of Modernizing Applications with Dell EMC VxBlock Systems,” a study of customer’s pre/post-CI experiences – published by IDC on Nov 7, 2017):

  • 61% lower cost of operations
  • 66% more efficient operations
  • 52% less time keeping the lights on
  • 34% faster application deployment
  • 99% less unplanned downtime
  • 8 Months pay back
  • 640% 5-year ROI

Look, people keep wanting to debate the role of CI and HCI, and what the future holds.  I get that.   I don’t think this is super-complicated:

  • My personal conviction, looking at a lot of data – is that both will continue for years, and BOTH will outgrow “build it yourself”.
  • HCI will of course out-grow CI, and HCI will be the “base” of IaaS/PaaS/CaaS clouds that depend on even more simplification/standardization and automation of lifecycke of the infrastructure layer.
  • At the same time – CI will be the “simple button” for consuming classic infrastructure stacks for apps that don’t land on an IaaS, but land on virtualized infrastructure.
  • We have refined, simplified and re-invented CI today – and I know the team isn’t stopping here, expect more.  We will keep using our experiences as the CI leader to keep pushing forward.

While I no longer lead the CI team (more on this in a bit – I’m keeping a little quiet) – I know they are awesome, and together with the HCI and Ready Solutions teams – they are in a great spot. 

On a personal note – I want to congratulate the whole Converged Infrastructure team at Dell EMC on the big day, the big launch – celebrate, and then keep pushing forward together to keep being the best!

Customers, partners (heck, competitors too!) what do YOU think?  What do you think about the new VxBlock 1000, and what do you think the team should do next?