I’m an MMA fan. The sport of mixed martial arts combines the multiple disciplines of wrestling, boxing, and jiu jitsu into one combat sport fought in an eight-sided cage called the octagon. These athletes are modern-day gladiators. Most come from a core background in one discipline (a collegiate wrestler, for example) and then have to develop secondary skills – boxing and jiu jitsu in this case – to really become professionally competitive in the sport.
I am also an entrepreneur, technologist and CEO. For me, announcements this past Tuesday in the cloud world are a metaphor for two highly trained MMA fighters stepping into the ring after a two-year training camp spent developing their secondary disciplines in preparation to battle for Cloud Service Provider (CSP) supremacy.
I am, of course, talking about Citrix’s acquisition of Cloud.com and VMware’s announcement of vSphere 5 as the foundation of its move toward a comprehensive cloud OS. To me, both announcements foreshadow three major shifts that will occur in the coming years in the cloud computing and CSP space:
1. A cloud OS will reign. The versions of Windows, Linux, Solaris, AIX, UX, etc., as we know them today will be rendered useless by the science of cloud computing. This will take some time, and perhaps elements of this layer will be repurposed, but both the hypervisor and the application will incrementally subsume functionality historically performed by the OS. We will see more intelligently pre-configured virtual machines (VMs) running workloads for increasingly intelligent applications pre-configured for a particular hypervisor.
We are already starting to see this with JEOS (Just Enough Operating System) VMs that intelligently adjust for attributes such as memory allocation through their interaction with the hypervisor. Dynamic resource allocation to VMs and the ability of the applications to drive such requests will represent the infrastructure of the future. This will be a core shift in the cloud computing space and will drive the orientation of the foundational levels of the cloud stack, Infrastructure as a Service and Platform as a Service.
This is like having a core base skill of wrestling for an MMA fighter. All other skills sets leverage the base. A case could be made that VMware is the better wrestler today given its heritage of innovation at the proprietary hypervisor and hypervisor service-provisioning level. On the other hand, it could be argued that Citrix’s acquisition of Cloud.com — which has enabling IaaS software IP that will undoubtedly provide the base to Citrix’s efforts to build out its own open source-based cloud OS (perhaps augmenting or even usurping its own Project Olympus efforts) — gives it the better wrestling skills.
Too close to call at this point. They have two different styles of wrestling.
2. The hybrid cloud will become the most important cloud. Most of the dialogue today is about which cloud is best for a particular enterprise when considering the choice of commodity clouds (Internet-facing web-scaling clouds such as Amazon Web Services), public clouds (multi-tenant clouds that can either be Internet- or private-access-based) or private clouds (essentially, dedicated single-tenant virtualized technology).
Key considerations are the obvious ones: security, performance, elasticity, etc. Greenfield applications generally have entirely different requirements than legacy back-office application migration (a space more commonly being referred to now as the enterprise cloud). It goes on and on.
The space is entirely over-marketed in a self-serving manner toward the strengths or away from the weaknesses of the CSPs doing the marketing. All of that will get sorted out in the next 18 to 24 months. Simply put, the best technologies for the required use case will win. If not, shame on the buyer.
As a CSP, the real juice in forward markets will be in defining and controlling the hybrid cloud. By hybrid cloud, I mean the heterogeneous solution of on-premise private clouds interacting with off-premise public clouds. The better the tools are to build on-premise private clouds, the more bursting will occur into off-premise public clouds. It will be a virtuous cycle.
Today, there are very few companies that functionally make money on both ends of that spectrum. I look at both the Citrix and VMware announcements as moves toward filling this future hybrid cloud opportunity. Citrix moves toward this in its acquisition of Cloud.com, whose software assets have been used to build such scaled private clouds as Zynga’s Z Cloud. VMware moves toward this simply by realizing it is no longer good enough to own the private cloud hypervisor market. It is now looking to develop true cloud-management products.
They both have pieces of the off-premise puzzle (and plenty of capital to develop it further) but neither is there yet. This is where this race will be. This area will require both skill and endurance and a tough chin, similar to what the art of boxing means to the MMA fighter. Given VMware’s massive share of the enterprise hypervisor market, it gets the nod as the better boxer at this point, but it’ll have to be wary of Citrix’s open source agility.
3. CSPs will need it all. It’s obvious that having a full cloud stack will matter, but it will need to be integrated in a manner that does not exist today and will broaden its definition. Having an enterprise-grade hypervisor in this fight is sort of like breathing: It will keep you alive, but won’t get your arm raised in victory.
Both VMware and Citrix have battle-tested hypervisors in ESXi/vSphere and XenServer, respectively. Both have data center assets and now IaaS and cloud OS capabilities. Both have management tools. Both are in their early stages with their PaaS offerings, with VMware’s recent launch of Cloud Foundry (interestingly, an open source PaaS) and Citrix’s OpenCloud, which leverages OpenStack.
Both also have a software portfolio, but this is where I think Citrix separates itself in a number of areas. One example is the network/WAN optimization for on-premise to off-premise connectivity noted above. The maturity of Citrix’s HDX technologies in conjunction with Xen Desktop have focused on solving the “last mile” of on-premise user experience with off-premise workload processing.
But all of these components must be interconnected. This is about a breadth of skill sets that work together in a “cloud spectrum” of sorts. This is similar to the breadth of specific technical skills that the MMA fighter must master in the art of jiu jitsu. At this stage, I give the nod to Citrix for having the better jiu jitsu.
VMware and Citrix are premium fighters
However, there are many other contenders in the division. The list is long. Don’t count out AWS with its impressively consistent feature release cycles. Rackspace has approximately $1 billion of dedicated hosting and managed services clients to direct to the virtues of OpenStack. Microsoft, IBM and others have well-known component solutions but have demonstrated a lack of agility, among other things, to enable the true spectrum. There are contenders in Red Hat, Joyent, Virtustream and many others.
There will be no shortage of opportunities for very smart component technologies. The major players will either have to develop these things or acquire them. Boxers must learn to grapple, kick, and employ submission techniques. Wrestlers must learn to strike effectively and attack from their guard. One-dimensional CSPs need to become effective three-dimensional CSPs in order to win a belt.
VMware and Citrix just look like the top contenders right now. Yesterday was the weigh-in for their title fight. But expect more days like last Tuesday.
Rodney Rogers is co-founder and CEO of Virtustream.
Image courtesy of Flickr user fightlaunch.