VMware Hybrid Cloud Strategy: Near Term Momentum but Icebergs Ahead

Premise

Wikibon Analysts:

David Floyer

Dave Vellante

The data center market is on fire and so is VMware. The company is consistently delivering on or exceeding the market’s expectations for financial performance.  Its execution seems to be incredibly crisp now. However Wikibon believes VMware’s current hybrid cloud strategy is incomplete. In our view the company needs to take aggressive steps to remain competitive in the long term. VMware needs to adopt what we refer to as a cloud-first development and functional release strategy; and invest its future cash flows to ensure its long-term position as king data center infrastructure software hill.

In addition, Wikibon believes VMware must both leverage and increase its commitment and contributions to open-source communities and third-party software. VMware must embrace outside sources as a vital part of bringing a cloud-first platform to market.

Wikibon believes that VMware needs very significant investments to ensure it continues to be competitive over the next decade. Dell is using $11 billion of VMware’s $13.3 billion short-term reserves to recapitalize Dell. If Dell inadvertently or deliberately distracts VMware from its mission, or fails to adequately fund initiatives, harm could be done to VMware’s value to customers and shareholders.

VMware Cloud strategy

VMware vCloud Air

VMware’s initial response to the public cloud was vCloud Air. This was a cloud service owned and operated by VMware, and built on vSphere and vCenter. It supported existing vSphere workloads and third-party applications, as well as new “born in the cloud” application development. It could be combined with existing on-premises vSphere to create a coherent hybrid cloud and a “like like” infrastructure between on-prem and public cloud. Essentially VMware envisioned a homogeneous hybrid cloud, which we would define as a combination of public and private cloud nodes with the identical platform running in all nodes. Such a system would offer integrated management support across the nodes. This is different from True Hybrid Cloud, which is defined in the True Hybrid Cloud section below.

vCloud Air offered dedicated cloud nodes and virtual private cloud services, as well as a range of recovery services. It was offered as a subscription service or as a “pay-by-the-drink” consumption service and on paper looked reasonable. The problem for VMware was many in VMware’s ecosystem didn’t want to build their public clouds on proprietary software and be subjected to the “V-Tax.” As well, VMware didn’t have the scale or market credibility to really make significant progress in the space.

After many “do-overs” and different attempts, like some other vendors’ (e.g. HP) efforts to directly compete in the public cloud  business, VMware capitulated. The company announced the sale of vCloud Air to OVH in 2016, and completed the sale in 2017. Instead, VMware turned to cloud providers as partners. IBM was selected as an early partner. In a major move, AWS was also chosen as a partner in 2016, and together built VMware Cloud on AWS.

VMware Cloud on AWS: A One-Way Trip to “Hotel Cloud-ifornia” or a Boon for VMware? 

VMware and AWS announced the joint partnership in October 2016 to create and market VMware Cloud on AWS. This is a scalable integrated cloud offering built on Amazon Elastic Compute Cloud (AWS EC2) “bare metal” infrastructure. The service is sold and maintained by VMware and allows enterprises to seamlessly migrate and extend their on-premises VMware vSphere-based environments to the AWS Cloud.

AWS got instant access to VMware’s 500,000 Customers. Many commentators predicted this agreement would result in a one-way ticket to VMware Cloud on AWS, followed rapidly by migration to native AWS services. This has not happened and won’t be as easy as some predict in our view. Conversion from one platform to another is a difficult, painful, costly, and risky strategy. Wikibon has always argued that conversion should be a last resort. VMware customers are happy with VMware tooling, especially around vSphere.

Nonetheless, customers have found that some workloads such as development can and should be moved to the public cloud, where a broader range of development services is available. In addition, recovery services to/from on-premises and public cloud offer very cost-effective and more flexible alternatives to a second datacenter.

Interestingly, some enterprises have initially developed new services on the VMware on AWS cloud, and then brought them back to run in production on prem. The reasons for migration of production back from the public cloud are different for different enterprises. One important reason is the necessity of avoiding the cost and increased latency overheads of moving data from on-premises to the cloud, when the original data is created from other applications running on-premises. Equally the reverse is true. If applications on-premises need real-time access to data generated by applications in the public cloud, applications can be moved to the VMware on AWS cloud and reduce the data movement costs and latency overheads.

This illustrates a strong principle that Wikibon has consistently maintained is a key requirement for any hybrid cloud enterprise strategy. Specifically, where possible, compute should be proximate to where data is created – i.e. compute should be moved to the data, and rarely data moved to the compute. Wikibon has done a number of studies on Edge computing, which strongly support the principle of processing data as closely as possible to its origin.

The other contributing factor to improved migration in both directions is the development of VMware SDDC and a True Private Cloud strategy. This is explained in the next section.

VMware True Private Cloud

A True Private Cloud is distinguished from a private cloud by the complete integration of all aspects of the offering, including hardware, software and other characteristics such as flexible pricing models. Equally important, the supplier of a True Private Cloud offers a single point of purchase, support, maintenance and upgrade. This is often referred to as a “single hand to shake, or  a single throat to choke.”

The key benefit of true private cloud is that it provides solutions close to the cost and agility characteristics of public cloud in an on-premises deployment, when business edicts dictate (e.g. security and latency requirements).

VMware SDDC (Software defined Data Center) allows VMware to run on x86 servers, without additional SAN or networking products being required. Essentially, VMware replaces traditional hardware function through sophisticated software and relies on off-the-shelf servers for compute. For example, the storage area network (SAN) is replaced by vSAN, and networking products are replaced by NSX. This allowed the birth of VxRail, a complete specification for a hyper-converged private cloud. VxRail is shipped prepackaged with all the software and hardware required to stand up True Private Clouds on premises. VMware is the leading in True Private Cloud, with 24% platform share (see Table 3 in the link).

Generally, software companies want to commoditize hardware. But it’s notable to point out that vSAN was hamstrung for some years for two primary reasons: 1) EMC (VMware’s initial owner) was focused on selling hardware-based SAN solutions to VMware customers and wanted to preserve its markets; and 2) VMware’s ecosystem comprised a “cartel” of networking and storage hardware suppliers (e.g. Cisco, NetApp and others) that were powerful partners and did much of the heavy lifting to integrate infrastructure and make it work more seamlessly on VMware. As such, VMware’s governance structure didn’t allow complete independence and VMware didn’t quite have the market power to alienate its ecosystem of hardware partners.

The result was a complex architecture to integrate the world of SANs to the VMware world. It’s generally believed VMware would have implemented vSAN much faster if it had been responsible for its own development priorities. The question is will Dell take the same posture or is Dell, as a leading supplier of servers, more partner friendly than was EMC?

True Hybrid Cloud

VMware Homogenous Hybrid Cloud is a number of nodes that all run VMware SDDC and vSphere. At least one of the nodes is almost always in a public cloud, and the other nodes distributed across on-premises nodes and/or other VMware public clouds. Many of the infra-hybrid cloud services are vSphere related. Examples would be VM migration from one node to another, fail-over and fail-back of nodes, etc. The nodes can be at different version levels of software. Each node is responsible for running and safeguarding the systems and data under their control. VMware is enhancing the management of services across all the nodes of a Homogeneous Hybrid Cloud.

The key advantage of Homogenous Hybrid Cloud is that software can be run where the data is located. This reduces data movement costs, and allows flexibility in designing system workflows.

However, there is one major disadvantage of the VMware Hybrid Cloud model. It is difficult to manage workloads across the network. Each node is mainly self-sufficient and independent. Updating functionality has to wait for new releases of software, which is a slow process. Figure 1 in the section below has more detail.

Emerging solutions have taken a different approach. They use what Wikibon has termed a “Cloud-first” architecture to develop a True Hybrid Cloud. This allows far faster development and adoption of functionality.

Wikibon believes that VMware needs to move to a Cloud-first architecture, as described in the next section.

Cloud-first True Hybrid Cloud Architectures

“Born-in-cloud” software has different characteristics than traditional data center technology. Figure 1 on the right shows the differences between a “Cloud-first” development and architectural model, and the Traditional Model. The middle column shows the impact of a Cloud-first approach.

Code is usually developed using micro-services. Improvements are made rapidly and incrementally to the cloud systems while the system is running. The objective is to improve the service built on the cloud software as rapidly as possible. These types of system architectures are referred to as “cloud-first “architectures, or as using a cloud-first model. All the major cloud services providers (Alibaba, FaceBook, Google, Tencent, etc., etc.) have used this model. AWS, in developing its Infrastructure-as-a-Service offering, follows the same cloud-first principles.

VMware has not been built using a “cloud-first approach. Improvements are made in new releases, rather than incremental upgrades. The differences with a Cloud-first approach are shown in the third column of Figure 1.

The newest entrants into the True Hybrid Cloud arena have adopted this cloud-first model. Microsoft Azure Stack has been developed this way. The initial software is developed in the cloud. Azure Stack is then populated with exactly the same software, and managed directly from the cloud. New versions of the software are pushed down rapidly to on-premise nodes. Very strong management of the on-premise platforms are fundamental.

Oracle Cloud at Customer has a similar cloud-first model, and is implemented further up the stack to include the database layer. Again, there is very strong management of the on-premises Oracle hardware. Wikibon expects that high-level applications such as Oracle eBusiness Suite will follow, using a SaaS model to provide the same code on premises or in the cloud.

AWS uses the Cloud-first model in the rapid development of AWS IaaS services. Wikibon expects that AWS will also use the Cloud-first model when it enhances its on-premises “Snowball” solutions to a wider set of hardware platforms.

The key advantage of the Cloud-first model is faster time to value for new functionality reaching users. In addition, it aids the development of central tools to manage all the complete True Hybrid Cloud instances. These tools use data from across all instances, and are based on advanced analytics and AI. They can be introduced to automate complex operational procedures across the whole True Hybrid Cloud instance.

Another advantage of a cloud-first model is that software from open source and ISVs can be introduced much faster. This is discussed in the next section.

VMware Ecosystem and OpenSource

Open source and small software companies are a critical source of innovation for digital cloud services. Google has been a leader in contributing and utilizing open source code and projects (e.g. Kubernetes).  IBM and Microsoft have implemented aggressive initiatives to foster and contribute to the development of open source code and projects. Much of the proprietary software being built is based on code that started as open source, with AWS leading the way in using the open-source code, if not in giving back, (an often-noted criticism of AWS).

VMware has started to embrace open source and started to contribute to the developer community. VMware needs to be at the forefront of new ideas to help implement modern application architectures and designs. New systems architectures based on PCIe and fast converged ethernet switching technologies are creating major new application opportunities, and open source software and developer projects will be an important asset for initiating and testing new ideas. The Gen-z initiative, which is expected to later in 2019, will again bring new opportunities for radically advanced middleware designs. VMware needs to be much closer to the hardware, and design software and hardware together. It cannot be bound by Dell in what hardware approach is taken.

Wikibon believes that VMware must significantly increase its commitment and contribution to the open-source communities, and embrace open source as a vital part of bringing a cloud-first platform to market. The developer community will increasingly be the selectors of cloud platforms for future high-value applications that only a couple of years ago were impossible to implement. Figure 1 shows the benefits of a Cloud-first model for implementing open-source and ISV software; being part of the project vs. always being in catch-up mode.

A key question is whether VMware has access to the funds necessary to develop a cloud-first platform. The next section discusses the issue.

Dell’s Recapitalization and it’s Impact on VMware

Dell technologies is a private company that acquired EMC for $67 billion in 2016. With it came EMC’s most valuable asset, 82% of VMware.

VMware is a publicly traded company, with a market  value of $62B (billion US dollars) as of August 25, 2018. Dell technologies owns 82% of VMware, at a valuation of $51B. The total market evaluation of private companies is difficult. A market consensus from a number of financial sources estimates the market value of Dell technologies (including VMware) to be in the order of $90B. This implies that the Dell VMware holdings are about 57% of the total.

The VMware Financials as of August 25, 2018 are shown in the Table 1 above. The results and growth show a very profitable, traditionally well-run company with solid growth and excellent profitability, with significant cash on hand of $8.1B. This cash has been recently growing by about one (1) billion dollars per quarter. The total “cash, cash equivalents, & short-term investments” in the VMware Q2 financial report is $13.3 billion, as shown in Table 1.

Dell and Silver Lake Partners, as VMware’s primary shareholders, are using the profitability of VMware to recapitalize Dell through a cash and stock buyout of the shareholders of VMware’s tracking stock (DVMT). As widely reported and included in Dell’s SEC filings, investors in DVMT can swap the VMware Class V tracking shares for either 1.3665 in Dell common shares or $109 in cash, with a ceiling of $9 billion that will be allocated on a pro-rata basis. Notably, the cash portion of this transaction will be financed by a special $11 billion dividend paid by VMWare, out of the VMware 13.3 Billion in cash, cash equivalents & short-term investments.

Is this the best use of VMware’s cash? Has the strategy of increasing VMware cash reserves had an impact on investment projects over the past two years? Clearly if VMware were a completely independent company the answer is “no” and “very probably” respectively to the two questions. But VMware is controlled by Dell and this presents risks to VMware’s long term dominance. Dell must play a careful balancing act between funding its corporate development efforts and allowing VMware to fund its future.

Wikibon believes that VMware needs very significant investments to ensure it continues to be competitive over the next decade, and to grow its cloud revenue by the percentages being achieved by AWS, Microsoft and many others. VMware has created the funds necessary. Dell will need to provision these funds for VMware cloud-first investment, and avoid using VMware as a piggy-bank.

Action Item

VMware’s value to customers and shareholders will be maximized if Dell does not distract VMware from its cloud-first hybrid cloud mission, and adequately funds VMware cloud-first and open source initiatives. Senior IT executives should obtain assurances from Dell and VMware senior executives that VMware is investing as aggressively as possible in cloud-first and open source initiatives.