EMC Buys VMware: Bridging the Server Storage Boundary ... Or a Bridge Too Far?

EMC recently acquired VMware (Palo Alto, CA; privately held, 360 employees, less than $100M in revenues) in a cash deal valued at $635M. VMware is the leading vendor of server virtualization software for 32-bit Intel-based servers (supporting Windows, Linux, and NetWare). VMware's growth has been explosive, with revenues doubling for each of the past several years.

META Trend: Through 2006, non-mainframe infrastructure consolidation will be driven by value-based portfolio management principles, but will be hampered by tool immaturity, premium-priced high-end servers, inflexible software pricing, and chargeback politics. Infrastructure co-location and networked storage consolidation will become standard by 2004/05. Best-practice infrastructure cost-efficiency will be defined by effective software asset and vendor management, support cost containment, and extension of server and storage refresh cycles. However, maturing server partitioning and integrated workload management will have minimal impact on staffing and total cost of ownership. When coupled with exorbitant high-end server pricing, these issues will inhibit "big box" consolidation.

This deal is indicative of the importance of two key industry trends: 1) the growing importance of an adaptive organization (AO); and 2) the increasing perception of hardware commoditization, which is largely being driven by the rising tide of Intel-based systems and n-tier application architectures.

The speed and the volatility of global markets are driving businesses to require that IT costs be more dynamically aligned with business demand. Indeed, nearly all major IT vendors are pushing some form of AO and IT infrastructure virtualization (e.g., IBM’s on demand, HP’s Adaptive Enterprise, Sun’s N1, Microsoft’s Dynamic Systems Initiative).

Although many AO concepts (e.g., alignment of IT with the business, variability to IT cost, leveraging of selective outsourcing, building of infrastructure resource pools) are not new, the aggregation of such concepts represents an overall philosophy change in how the transformation can be achieved. While IT organizations (ITOs) are still striving to understand the breadth of the concepts, much of the message continues to be driven by vendor marketing and has only begun to get the attention of business-unit managers. Although great technical strides will continue to extend virtualization and application performance, the lack of flexible interoperable or pricing model standards, resource utilization, and chargeback models will constrain virtualization adoption and complicate user total-cost-of-ownership evaluations.

Moreover, the broader AO vision encompasses many different technologies (e.g., blade devices, SANs, virtualization, automation, self-healing, Web services, service-oriented architecture), sourcing options (e.g., insource versus outsource), and business structures (e.g., variable infrastructure pricing). We do not anticipate (nor do we recommend) that organizations will select a single “adaptive” vendor. Rather, they should choose to fill a specific need and select vendors to meet that need, with an eye on how that technology fits the overall vision of being adaptive.

EMC is clearly attempting to position itself as a key player in this adaptive space with its ILM (information life-cycle management) initiative. To the extent that AO adoption will be characterized as incremental, heterogeneous, and federated - rather than single source, monolithic, and full scale - this trend will benefit EMC’s approach. Moreover, EMC is characterizing the VMware acquisition as part of its strategic shift away from being primarily a storage hardware vendor to becoming more of a software vendor. The VMware deal closely follows the recent acquisitions of Legato (July, $1.3B stock) and Documentum (October, $1.8B stock). Indeed, we expect software sales to drive more than half of EMC’s 2004 revenues. With about $6B in cash, additional acquisitions by EMC would not be unexpected. However, these software outfit acquisitions are not naturally aligned by either technology or technology domain. Although EMC has proven to be capable within a storage-centric domain, EMC is relatively inexperienced in terms of a broader adaptive infrastructure stack, and we expect significant integration challenges.

We believe VMware has resisted numerous acquisition offers. VMware is “server vendor neutral” and did not wish to become part of any particular system vendor. Moreover, Microsoft’s February deal to acquire many of the key products and staff of Connectix (VMware’s principal competitor) was in response to VMware also wanting to remain “operating system agnostic.” However, we believe more than 80% of VMware’s revenues are Windows-based and, perhaps somewhat ironically, Microsoft had been VMware’s largest end user.

There is very strong market momentum for VMware within META Group’s Global 2000 client base. VMware has become the de facto standard for Intel server virtualization. Our research indicates that, though ITOs are initially attracted to VMware as part of a server consolidation play, VMware is especially popular in both development/test and production environments (roughly 50/50). While VMware runs on larger systems, VMware provides only limited virtual machine SMP support (up to two virtual CPUs with its ESX product), though this covers up to 80% of Intel workloads. We anticipate four-way support in 2004. This is in contrast to Unix-based partitioning capabilities (e.g., Sun Domains, HP vPar and nPar, IBM LPAR), which generally provide SMP partition support (with fractional CPU support expected in 2004). In addition, one area with significant virtualization overhead (more than 20%) is with I/O-bound workloads, though clearly EMC has considerable expertise with SAN-based I/O (albeit less so for networking).

However, VMware has also done an excellent job in diversifying its value proposition. It is often used to facilitate operating system migration (e.g., hosting NT 4 and Windows 2000, Windows Server 2003, or all three) and rapid-development and test-system provisioning, as well as in increasingly creative high-availability and disaster recovery alternatives. Overall, we believe VMware continues to have a 12- to 18-month lead over Microsoft (Connectix). Future offerings look promising with VMware’s server virtualization and EMC’s storage management software capabilities, coupled with Documentum’s policy-driven data management. The key challenge for both EMC and VMware will be their collective capability to virtualize Intel I/O, effectively allowing I/O to be separated from virtual partitions (not physically assigned), and to “float,” providing more efficient I/O capabilities.

Although VMware has become dominant, its long-term future was being threatened by Microsoft’s own Virtual Server initiative, which will be more completely integrated within Microsoft’s Dynamic Systems Initiative in the Longhorn time frame (and in feature packs prior to that). Fortunately, Intel recently announced its longer-term Vanderpool technology initiative to increase hardware-level processor virtualization. Intel will be significantly expanding the number of CPU cores per chip and threads per core. We believe Vanderpool will appreciably reduce virtual machine overhead and enable a more level playing field for software vendors.

Although VMware will become a key component of EMC’s broader software-driven future, we believe EMC’s ownership may hinder VMware’s future relationships with IBM and HP, once Microsoft’s technology matures. In addition, current Linux deployments rarely include EMC. However, Dell will continue to benefit, given its very successful EMC partnership.

Bottom Line: We believe ITOs will see little change with VMware under EMC’s ownership for at least 12-18 months, as EMC works its way through significant and complex product rationalization across the breadth of its recent acquisitions. Longer term, EMC will begin to challenge the adaptive infrastructure initiatives of broader industry players such as IBM, HP, and others - all of which have a sizable head start. Ultimately, as a software vendor with no server but a storage hardware agenda, EMC will have to both partner/integrate and compete with these vendors.

Business Impact: An adaptive organization is key to aligning IT with business variability. ITOs and business leaders must quantify both the magnitude and timing of the costs and benefits that will deliver value. Rather than wait for full AO deployment, each phase in the AO evolution should be justified by its own “standalone” ROI.

META Group originally published this article on 18 February 2004.