On March 9th, 2009, EMC held its financial analysts meeting in Boston. The highlight of the event was VMware's CEO Paul Maritz putting forth EMC's and VMware's vision of how virtualization will transform the data center and what role EMC will play.
EMC's/VMware's intention is to invest in the development of a platform to enable customers and partners to build private clouds that essentially, according to Maritz are "Software Mainframes." EMC laid out its strategy of how it intends to compete long term with the likes of Microsoft, IBM, HP, Google and Amazon.
EMC's three main pillars of value are:
- Choice - vCloud to enable cloud federation that supports a rent or buy decision;
- Control - through VDC-OS, the cloud foundation and VMware's virtual data center operating system;
- Efficiency - by making the desktop a service and making the user, not the device, the focal point of utility.
EMC's vision for the cloud is to allow customers to control and secure data/information and take advantage of the economics of elastic infrastructure. Rather than compete by placing megascale 100+ megawatt data centers on the Columbia River, EMC and VMware plan to create a federated model, virtualizing infrastructure, including partnering with Cisco and other ecosystem members to compete for cloud applications.
Maritz described VMware's evolution from a client hypervisor company in 1999 to a server hypervisor to cooperating hypervisors where internal and external clouds are encapsulated and managed by 'cloud operators.' This vision assumes we'll be moving away from a world where applications dictate / control the infrastructure to one where the cloud, based on standard building blocks, is deployed as an architecture and a way to buy acquire and provision resources.
The standard components of the cloud include:
- A management layer to provide SLA's
- Policies for security, compliance and governance
- Software - to scale and provide high availability as a single logical resource
- Standard x86 hardware
Importantly, this vision accommodates both new and existing applications, where complexity is encapsulated in a 'black box.' VMware has 2,000 engineers working on this vision to build the 21st century mainframe. In this scenario, applications think they're talking to a single giant computer or a massive switch with infinite and elastic cascading fault tolerance.
Maritz clearly indicated that it is VMware's intention to support performance and availability of ANY type of application. EMC storage fits in this vision with both traditional and cloud-based storage (ATMOS) plugging into the model. Cisco is perhaps EMC's most important partner providing distributed switching capability and intelligence into the fabric.
A key aspect of EMC's vision is a policy-based system with well-defined standards that can enable service providers to talk with each other and achieve economies of scale through federated infrastructure.
The bottom line goal stated by Tucci and Maritz was that they wanted to get customers out of the plumbing business and into the business of building applications.
Wikibon believes the software mainframe metaphor is extremely powerful and VMware's goal to support any application is ambitious and bold. We note two factors that stand out about a giant, mainframe-like computer:
- The sharing of resources over geographically distributed distances (albeit with the mainframe at synchronous distances);
- A system that you never have to restart.
The latter is a substantive challenge. VMware/EMC must earn the right to compete for supporting any and all applications. Injecting a major software layer into the data center environment and proving that layer to be as reliable as a mainframe will take time and considerable effort.
In addition we make the following observation. EMC/VMware are lining up many if not all of the components to compete for the cloud, including:
- Grid consolidation (e.g. the IT infrastructure),
- Ubiquitous fast networking,
- Support for standard rich clients,
- Open source (e.g. the LAMP stack),
- Mega scale data centers.
The last, data-center-scale, EMC plans to achieve via partnerships and a federated model. Maritz in his remarks cited numerous cloud computing services providers in EMC's/VMware's ecosystem. The companies are betting that by providing a mechanism to connect flexibly to heterogeneous service providers, they will be able to realize and match the megaplex datacenters of Google, Amazon, Microsoft, and others.
If EMC pulls this off, who wins? Customers, Cisco, service providers, ecosystem players and of course, EMC. Who are the target competitors? Microsoft, Google, Amazon, IBM, HP, Sun and possibly Dell. Of these, Microsoft has the most to lose, as VMware is attempting to usurp Microsoft's platform position.
Can EMC make money in the cloud? If the company pulls this off, it will make plenty of money selling software, services, cloud storage and conventional storage.
Can EMC/VMware pull this off? Yes. VMware is in the catbird seat and is gaining momentum. But it will not only take solid execution but also an ecosystem and developer community that lines up behind EMC. They're late to the cloud platform game but the game hasn't started yet.
Action Item: Customers should begin thinking about reference architectures that blend internal and external clouds and unite them under a so-called private cloud infrastructure. Key to this approach will be governance, compliance and a clear understanding of policy tradeoffs. As always, data classification is a starting point to matching service level requirements with available market options.