(just to be clear, I was giving interviews to journalists and most certainly not delivering keynotes!).
READER WARNING: this is a long post – get yourself a cup of tea, some digestive biscuits and clear your diary!
The two keynotes I caught were those from VMware heavyweights, Pat Gelsinger, CEO and Carl Eschenbach, President & COO. I really admire VMware’s humility in resisting the temptation to grandstand their obvious dominate market share in virtualization and instead to concentrate on all they are doing for customers. Theirs is an understated approach – especially given a captive, and seemingly ‘captivated’ audience of 8,500 – and those keynotes were all the better for it. Both gentlemen ooze quality and clarity and are extremely competent in their delivery. Of course they were supported on stage by customers, partners and senior technical execs from VMware , but it all was exceedingly well done. No showboating or aggrandizing. It was on a relatively modestly presented stage and they concentrated on the substance rather than power plays.
The major themes remain consistent from last year, but they have been fleshed out much more thoroughly. IT as a Service (ITaaS) remains the most overarching theme of all, and with its sub-themes of Cloud, End User Computing and Software Defined Data Centers, it might be considered as more an instantiation of their (IMHO) world domination plans 😉 that I wrote about previously.
The keynote’s title – ‘The Next Generation of IT’ – may seem grandiose and hyperbolic but the vision is credible. Pat stated that Gartner estimates that for a typical CIO only 30% of his or her budget is spent on innovation. From their own customer surveys VMware say that for CIOs who have invested in their technology that figure rises to 40%. And, as Pat outlined, the short-term goal of VMware is to enable customers to implement more services-based solutions and drive that innovation percentage up to 50%. Rolling out applications at the ‘speed of business’ and routinely deploying any workload anywhere in any architectural paradigm should lead to this ability to increase the innovation investment.
From the Cloud side, IDC recently ranked VMware #1 in Cloud Management and they have expanded vCenter Operations Suite with enhanced log management capabilities. The VHCS (VMware’s Hybrid Cloud Service) has come along rapidly since announcement earlier this year and Gelsinger announced the general availability of the service in Europe.
End User Computing, which my colleagues inform me was virtually an afterthought six weeks ago at VMWorld in San Francisco, was prominent in coverage in Pat’s keynote here. This was mostly because during the keynote, Sanjay Poonen (recently joining from SAP to be the EVP and GM for End User Computing), came on stage to announce the acquisition of Desktone. Desktone are the pioneers of Desktop-as-a-Service (DaaS) and adding this capability to the EUC suite, which already has a strong VDI offering in VMware View, really put the spotlight back on the end user portfolio. This acquisition is something that we at NetApp keenly support. Desktone is a very important NetApp partner and a leading FlexPod CVD. Gary Hocking, our CTO for our Service Provider business has written an excellent analysis and rationale for us here. It is well worth the read.
And finally, the third and possibly the most significant theme (if judged on airtime, combined from both keynotes) remains that of the Software Defined data Center (SDDC). The original goal, a few years ago was to deliver the ultimate, über “data center operating system’ of the future, but that has morphed into a definition of “data center” that goes beyond the limitations suggested by the word ‘center’ to encompass a range of deployment methodologies from on-prem, private cloud, public cloud and a conflagration of all three – the ‘hybrid cloud’. Delivery of an SDDC requires (in the world according to VMware anyway) the virtualization of all IT. Nothing should remain bare metal, nothing needs to be tied to hardware. Encapsulation and abstraction are the watchwords. On top of this, everything needs to be automated, up and down the IT stack. For VMware, an SDDC predicates virtualization of compute, network and storage, and the replacement of management tools with automation.
Total virtualization of compute is not 100% achieved yet. There are some stubborn apps that are resisting the treatment – or maybe it is that it is the CIO and IT staff who are resisting implementing. Probably a bit of both. Some of the big ticket items like ERP and core financials are taking time to get on the virtualization bandwagon within IT departments. Virtualizing business critical applications (such as SAP) remains very much a mission for VMware, together with its partners like NetApp.
Latest point versions of vSphere and VCloud Suite were announced with both focusing on significant improvements in such areas as the numbers of cores supported, the number of virtual machines and the number of virtual CPUs. Basically faster and better (can’t say whether cheaper though).
So, let’s assume that server virtualization is “more or less” done. Next up the stack is networking, and it seems to me that this conference has been the EMEA coming out party for VMware’s NSX. NSX is the network virtualization technology based on Nicira, the VMware made back in July for lofty sum of $1.2bn. Software Defined Networking (SDN) will be a core component of SDDC. Today organizations are heavily dependent on their routers and switches. Although I don’t work in the networking space even I know that these hardware elements are slow to provision, unwieldy and costly.
VMware aim to do to this gear what it did to servers – basically emasculate and reduce.
As Carl Eschenbach stated in his keynote, 70% of packet traffic is VM to VM and the dependency to drive the networking to the switch from one VM and then out and back is an overhead too far. Putting that capability straight into the VM will reduce the network traffic dramatically and of course the cost. So, you won’t need so many switches and routers and the ones you have will not need to all be the equivalent of an Oxford professor. This is the goal of NSX. Essentially it is delivering a “Network Hypervisor” akin to the ESX for servers.
So won’t this change the dynamics of the network industry like it did for the server industry? Well, yes, every indication is that is exactly what will happen. Gelsinger shared a logo-heavy slide with over 30 partners who are all supporting this new ecosystem. One very obvious and notable absentee off the list (and in fact in both talks I heard) was Cisco. The networking giant has potentially a lot to lose if NSX becomes the defacto virtualized networking platform. Obviously Cisco is not sitting idly by to watch this happen. In fact, their answer to network virtualization will debut in 3 weeks time – Insieme. My reading of it suggests that it will be architecturally quite different to NSX. I have no doubt that it will be positioned by Cisco from a marketing viewpoint so that it is not seen as directly competitive to NSX (most likely by emphasising the application centricity of it (addressing higher layers)), but it is publically described as focused on reducing the number of routers and switches and building more intelligence into software through abstraction.
Sounds like the same message to me!
Cisco and VMware remain committed partners – the former is a very successful OEM of the latter and of course they are partners in the VCE company. However it is clear that they may start stepping on each other’s toes.
The final element in the ‘virtualize all IT’ mission is that of storage. Storage is complex. On the surface, storing data on media, whether spinning disks, tapes or the newer SDDs, sounds straightforward. However, anyone in the IT industry who has had any dealing with storage knows how abstruse and complicated it can be. NetApp knows this – it has become a $6bn+ business on the back of simplifying it. Virtualizing storage – or ‘attempting to virtualize storage’ has been in play for many years. EMC came out with a product called Invista several years back and attempted to build virtualization as a separate software layer. It had limited impact and was quietly shelved. Other products for storage virtualization have also emerged from Hopkinton but again have followed a software layer architectural model. NetApp had one big advantage over its competitors – it had a single, highly scalable storage OS, Data ONTAP which has been incrementally enhanced over a 20 year period to provide all the functionality you could want from a storage OS – from snapshots to dedupe to compression to thin provisioning to cloning to ‘yadayada’, as my transatlantic cousins say. Data ONTAP is the most widely deployed storage OS in the world – and for very good reason. It is the Rolls Royce at a Mercedes price. With this single architecture adding in virtualization capabilities was not a vision but a reality – and happened over six or seven years ago. The idea of pooling all storage into logical pools and abstracting them from the underlying physical hardware was a feature back then. Since that time NetApp has added bigger pools, cross cluster capability and the industry’e leading virtualization solution through SVMs (storage virtual machines).
So the question is – what will VMware do to “virtualize” storage? Well, they are positioning vSAN (Virtual SAN) as their play at the storage layer. However I am sure they would be the first to admit – vSAN is for very specific use cases. Firstly it is DAS focused. It is aimed at the storage which sits in servers. Given the rise of blade servers, which typically do not have attached storage, that is a pretty limited market. Secondly, it is VMware-only. Thirdly it is limited in scalability compared to the independent storage vendors and fourthly it is still in beta. To be very fair to VMware they do make it clear that VSAN aimed at very specific use cases – branch offices, tier 3 storage and potentially VDI. In addition they reiterate their desire to continue to work with the storage ecosystem as the full promise of the SDDC is becoming realized. Unlike server virtualization (which they pretty much invented) or network virtualization (which they are bringing to market after acquisitions), storage virtualization has been around for quite a while – and the storage vendors are “all over it” – to use a figure of speech. VMware has always be an ecosystem-centric company, unlike say an Apple, and will remain so.
But here is the rub – and this brings me back to the 4 interviews I had with journalists from different countries within EMEA here in Barcelona – all 4 asked me whether NetApp felt that VMware are “stepping on our turf” with vSAN. For those that understand the VMware strategy, the various offerings and the ecosystem, the answer is a resounding ‘NO’. However, as I said earlier, ‘the devil is in the detail’ nd it is now behoven on us all to make the positioning clear to our mutual 40,000+ customers.
So to bring us full circle – yes, VMware is still on its course for world domination but NetApp is right by its side!!