The new blog contains highlights from our most recent report Ripple Effects from Virtualization. Please check out our new blog location to read about how server virtualization will disrupt the storage, database, and middleware markets while creating large new opportunities in business continuity and cloud computing.

The report is based on feedback by 200 IT customers on vendors like VMWare, Citrix, Microsoft, Oracle, IBM, Symantec, CA, BMC, NetApp, EMC, DELL, and HP, and is also available here.

The recent posts about the economics of cloud computing between Nick Carr and Tim O’Reilly (here and here) and the panel at this week’s Web 2.0 have created a lot of buzz. The central question is of great consequence: will the emerging “cloud” operating system generate the monopoly rents and industry control that Microsoft enjoyed with Windows? For the sake of argument, let’s assume VMware is leading in private enterprise clouds and that for now Amazon leads public ones with Google, Yahoo, Rackspace, and Microsoft as contenders. It seems likely so far that Microsoft will have a significant advantage in private clouds, though not to the same extent as with Windows. Public clouds seem years further out, so they’re harder to handicap.

But at the center of the argument is whether dominance in either variety will come via Web 2.0 style “harnessing collective intelligence” or the more traditional “network effects.”  I believe we will see Microsoft emerge as a leader in private clouds in 3-5 years and it will be on account of the more traditional network effects.  Market share won’t accrue to the leader by virtue of capturing more information every time someone uses the cloud.  Instead, it will accumulate the way Windows steadily accumulated application and device support over time.

The two critical success factors in cloud computing: virtualizing the hardware and managing the software components

There are two key assets to leverage success in this market. VMware and Microsoft are likely to share self-reinforcing leadership to the first. That is the ability to make a sprawling and heterogeneous collection of servers, storage, and networking look like a single machine. Through automation interfaces, this capability dramatically changes the economics of administering data centers. The other key asset to leverage for success in this market is the ability to combine infrastructure and applications management. That is the critical requirement for turning an IT operation into a private cloud that can deliver rock solid online services.

Leveraging network effects to make the hardware infrastructure look like one big machine

As Tim explains, the cloud has multiple layers. But the bottom-facing utility layer sits on the hardware infrastructure and uses this generation’s equivalent of device drivers to make it look like a uniform pool of resources to the software above this layer. This is VMware’s strength by virtue of its lead in virtualization. Market share begets network effects at this layer in terms of device support, such as the widest range of storage devices and access to all their unique features. However, over time, it’s hard to imagine how Microsoft will fail to leverage its Windows Server and associated Hyper-V unit volume to achieve similar device coverage. So although the utility layer intrinsically has low value add, vendor concentration in private clouds will probably preserve prices and margins to some degree.

Leveraging network effects to deliver end to end application service management

In order to be able to deliver end to end online services, the upward-facing cloud software layer has to orchestrate and manage an untold number of application components, many from third parties, some from corporate developers. VMware has some leading-edge management technology that automatically wraps around applications (courtesy of the B-hive acquisition). But as long as commercial and corporate developers primarily target Windows as their application deployment platform, Windows will have a self-reinforcing advantage relative to VMware.

Microsoft’s self-reinforcing advantages are twofold, building on its leadership both as a deployment and development platform.

1. First, since it accounts for roughly 80% of X86 server unit shipments, software developers of just about any stripe have to do at least some work to wrap Microsoft management tools around their applications.
   
2. Second, because Microsoft is the leading provider of development tools on Windows, it will be able to capture even more management information about the subset of applications that use their tools. Even if both of these structural advantages haven’t yet been exploited by Microsoft, it’s hard to see that situation lasting as they roll out their Dynamic Systems Initiative.

VMware’s best potential for upside in this market is threefold.

1. First, the extent to which commercial or corporate software developers use development platforms that are independent of Windows likely means neither company has an advantage in collecting management information. Examples of these platforms include J2EE, PHP, Ruby On Rails, Spring, and Hibernate. Some of these platforms don’t require any conventional operating system.
2. Second, the extent to which VMware proliferates its platform ubiquitously, it may have more of an advantage in managing the infrastructure it sits on and the applications that sit on it.
3. Third, virtualized environments make it possible to deploy applications in “appliances” that include all the bits required to run, including operating system, middleware, and multiple application components. Today, all these appliances are deployed using Linux because of Windows licensing restrictions. If appliances take off independent of Windows, that would help VMware tilt the platform competition in its favor.

Public clouds are likely to be more diverse, like a set of services, but with the core or anchor services having similar economics

Windows Azure has all the economic characteristics of a private cloud – masking the infrastructure, managing the application services – but with the margin-depressing overhead of tightly-integrated data centers. It will clearly have higher-level services like SQL Services, Exchange, and .NET, but it will be easy to integrate premise-based software as well as third-party services such as Salesforce.com’s Force.com or Google Adwords.  In other words, if Microsoft delivers on its promise, it will be an anchor platform at a higher value, higher margin layer than Amazon, but with bridges to other services.  Considering how it appears to build on Microsoft’s on-premise software tools and interfaces, it appears likely to have a leading market share.

The public’s excitement though around cloud computing overrates what is happening and under-estimates the dramatic impact that internal and external clouds will have in the medium term.

Hopes too high on pace of adoption

Our own research with customers indicates that businesses are slow to adopt the public cloud (outside of SaaS of course).  Top concerns are data protection (security, privacy) and lock-in to one cloud provider (lack of standards).  Early enterprise adopters are those with spiky traffic (e.g., media and entertainment) where the cloud provides low baseline cost with infinite scalability.

Our survey also indicates that enterprises are building private internal ‘clouds’ right now which are enabled through virtualization (see prior post just below).

We have heard repeatedly that joining these with external clouds is a few years out, though we are encourged that it’s on VMWare’s roadmap.

Mark Stahlman of Gartner is quoted in the article that “hardware always wins when new demand for computing is uncovered.”  While we agree that lower cost and complexity will drive new applications for computing, we see significant disruption to the IT industry as we know it once cloud computing gathers pace:

1. Server and storage vendors will face lower margins and a period of lower growth.  Customers move from ‘just in case’ forward buying to a ‘just in time’ capacity on tap model, which is a roadblock to unit growth.  Moreover, concentration of buying power in the hands of a handful of big clouds is a negative for margins, witness the recent Amazon – EMC deal.  We are not confident that the extra demand which Stahlman sees will compensate for that fairly tangible downside
2. Cloud technologies in the enterprise, growing out of virtualization and service-oriented management, promise at least an order of magnitude greater administrative productivity.  Today, one IT administrator typically manages 20-30 physical servers.  By contrast, these emerging private clouds can manage far more sophisticated and scalable application services without growing staff.  At the extreme, public clouds like Microsoft’s and Google’s leverage one administrator across several thousand servers.
3. Cloud computing will not provide material profit upside for the providers of large clouds anytime soon (e.g., Microsoft, Google, Amazon.  Rackspace is the only pure-play in the group.)  Cloud provision will be a commodity business except for niche clouds, e.g., ’high security’ for regulated industries or government customers where you need to prove or restrict the physical location of the data
4. While Windows may appear to be consigned to the history books today, Microsoft will attempt to change that with its announcements at the Professional Developer Conference next week.  We expect them to create a new set of tightly integrated online services that resemble Tim O’Reilly’s much talked about Internet Operating System.  It is likely to deliver not only dramatically higher administrative productivity for their internal operations, but dramatically higher developer productivity than any of the competing cloud platforms.  Microsoft shouldn’t be counted out yet.

By George Gilbert and Juergen Urbanski

We’ve talked to a fair number of VMware customers and investors over the past few weeks.  In the process, we’ve repeatedly been asked whether VMWare is done with its phase of hyper-growth.  While it isn’t likely to grow anywhere near triple digits again, it is likely to grow into a strategic platform provider for both data centers and desktops, though this will require solid execution in a tough macro environment.  Its opportunity comes from its chance to both expand and disrupt a series of large adjacent markets.  The ripple effects of this sea change in computing will also affect many markets which VMware has no plans to compete in, though that will be fodder for future posts.  (Disclosure: the authors own shares in VMware)

VMware’s biggest near-term challenge is that it over-sold both units and high levels of functionality with their enterprise license agreements.  These ELA’s were an attempt to encourage customers to deploy more virtual servers with richer functionality ahead of Microsoft’s entry into the market this past summer.  While this may have had some success in making adoption of Microsoft technology more challenging in some accounts, it has actually had unintended side effects.  It left VMware competing with its own inventory of licenses already on the shelves of its customers.  While VMWare works its way out of that near-term hole, some have lost sight of the bigger picture opportunity.

The Many Waves of Virtualization Technology That Start at the Server

Once VMware, followed by Citrix and Microsoft, inserted a thin layer of software between the server hardware and all the operating system and application software that now sits above it, a whole range of value creation opportunities opened up.

Everyone knows the story about the server consolidation opportunity.  The biggest misconception though is that the high growth story is done because Microsoft has commoditized server virtualization technology.  What is really going on is that there are multiple waves of functionality continuing to penetrate the installed base of servers.  Roughly in descending order of immediacy, these waves include server consolidation, business continuity, desktop virtualization, data center automation, and cloud virtualization.  These new waves greatly expand the addressable market for server virtualization because VMware is democratizing functionality that often exists in other markets at much higher cost and complexity.

Conventional wisdom doesn’t seem to take into account that each successive wave can reach existing and/or additional servers at price points that are anything but commodity-level.  The business continuity wave is already upon us.  In particular, high availability and disaster recovery are building on or replacing the original server consolidation value proposition just as Microsoft is entering with the previous wave.

The waves that are somewhat further out are desktop virtualization, which is actually managed by servers, data center automation, which turns internal IT infrastructure into a private cloud of resources and application services, and federation with public clouds like those that an Amazon or an IBM might run.

Just because VMware plans to deliver these new waves of functionality doesn’t guarantee it success.  In fact, the competitive dynamics of each wave are likely to be dramatically different.

The Server Consolidation Wave is Mostly Well Understood

The biggest misconception in this phase of market development is that Microsoft has driven price points per physical server (or per two sockets in VMware’s case) toward zero and that the number of workloads consolidated per physical server continues to climb with improving server price performance.  The reality is that ease of managing planned downtime via live VMotion remains a differentiator for VMware for now.  In addition, the consolidation ratio of workloads per physical server is trending down, not up, raising the price per workload.

Some estimates put the total percentages of physical servers that have been consolidated in the mid-teens and the total percentages of workloads that have been consolidated at least in the mid-twenties.  These numbers need more analysis because the consolidation ratio for early workloads, particularly test & development and tier 3 applications such as directory servers, DNS servers, and file & print servers, have supposedly ranged from 10 : 1 to as high as 20 : 1.  But this consolidation ratio is not continuing to climb, and hence the price per workload is not continuing to drop, as server price / performance continues to improve.  Customer plans are different.  Those we’ve talked to indicate that as the applications become increasingly sensitive to downtime, customers become increasingly cautious about stacking too many workloads on a single server.  They are starting with tier 2 business critical applications and extending that to tier 1 mission critical and customer-facing applications.  However, the consolidation ratios that we are hearing about are closer to 3 : 1 or 6 : 1.  This may increase over time as customers get more comfortable with measuring and managing utilization rates on these more performance-sensitive applications and as the hypervisor “performance tax”, currently 10-20%, continues to diminish.  But for now, the price per workload is going up, not down.

Figure 1:

Average 4-CPU Oracle database performance requirement is 1/80th of VMware ESX capacity, leaving plenty of room for consolidation.  Data based on actual analysis of 700,000 VMware customer servers.  Source: VMWare

In terms of where in particular Microsoft is gaining share, Hyper-V is proving attractive to customers migrating to Windows Server 2008 who need a “compatibility box” for hosting older versions of server applications.  Specifically, those who need to bring along Microsoft server applications running on previous versions of Windows find Hyper-V hosted on Windows Server 2008 attractive.  It’s not clear that Citrix has the channel to reach server infrastructure buyers generally, but they do have the most natural claim to consolidate the 1 million servers running what used to be called Citrix Presentation Server.

The Business Continuity Wave is Imminent as the Next Growth Driver

At a meeting for Wall Street analysts at VMworld, CEO Paul Maritz said the company had two years to come up with something to follow the server consolidation opportunity.  The thought that the original growth driver might be tailing off and that the company hadn’t identified the next leg of its growth strategy spooked investors.  The market promptly punished the stock the following day somewhere between 15-20%.  The event highlighted that consensus opinion doesn’t even realize that a follow-on wave is imminent.  And post-Diane Greene, the company is now over cautious about hyping its opportunities.

VMware has a rapidly maturing set of products for business continuity that builds on and extends the server consolidation value proposition.  The percentage of workloads covered by high availability (HA) and disaster recovery (DR) in a pre-virtual world is low because of the cost and complexity of implementation and operation.  Today, somewhere between 10-20% of workloads are covered by HA and even fewer by DR.  In a virtual world, that cost and complexity goes down dramatically, and customer deployment plans appear to be for 2-5x higher penetration rates than the pre-virtual world.  Current VMware list prices for this functionality are 3x the level for server consolidation functionality, or roughly $3K per dual socket.  Even if street prices drop as volumes go up and despite ever richer functionality, this opportunity appears bigger than the server consolidation wave.

Figure 2:

Today, application-specific clustering from Microsoft, Oracle, and IBM provide application fault tolerance but with complex setup and maintenance, severely limiting the applicability of HA functionality.  VMware on the other hand provides failover and restart upon hardware failure with minimal to no administrative overhead.  Citrix with Marathon Technologies provides fault tolerant application failover.  Source: VMWare

While potentially bigger, the business continuity wave is unlikely to take off at the same rate as server consolidation.  For one, the ROI is not as easily measurable.  In addition, VMware, Citrix, and their channel partners have to step beyond the traditional server infrastructure administrator and address the application owners and administrators.  This is a new buyer for both.  Citrix is the first to admit their ambitions on the server are limited because of their channel’s limited reach with this buyer.  For VMware, however, their opportunity starts with the voluminous number of previously consolidated servers.  Because many of these have so many workloads on them, customers now consider them too important to fail.  This is the entry point for HA deployments.

For HA, the principal competition comes from the existing server application and operating system vendors that have product-specific approaches such as Windows Cluster Server or Oracle Real Application Clusters (RAC).  These legacy solutions tend to be very expensive and complex, partly because administrators have to maintain replicas of servers in the cluster for the entire hardware and software stacks across configuration, patch, and version changes for each and every layer.  VMware and Citrix, by contrast, treat the whole software stack as a file that can move between servers, vastly simplifying maintenance and operation.  It is likely only a matter of time before Microsoft is compelled to adopt this approach as well.  Today VMware only protects against hardware failure, unlike Citrix – in conjunction with Marathon Technologies – and the server software vendors, who protect against software failures as well.

But VMware is building application awareness into its product and plans to offer full software-level fault-tolerance to applications.  At the same time, it will enable its ISV partners to do the same for their products as an alternative offering.  Customers regard virtualization-based application fault-tolerance as the next killer feature to follow live migration.  HA, where the recovery time for the app and recovery points for the data are measured in minutes, or FT, where both are instantaneous, are likely to become standard service levels.  Now that applications have many tiers or components, the availability of one is dependent on the availability of all.  It’s no different from stacking 10 workloads on a single server and then deciding the server has become too important to fail.  Customers talk about applying HA or FT to just about all physical servers and their workloads now that they are becoming so inexpensive and simple to deploy and maintain.  For example, Oracle’s Real Application Clusters cost an additional $25k per socket at list and are so fragile that customers for the most part only deploy them in 2 server FT clusters, not for scalability.  The mid-point of VMware’s HA functionality today, though not yet application-level FT, is roughly $3K per socket.  And the cost of ownership is roughly an order of magnitude lower: administrators can manage clusters with an order of magnitude more servers.

The Data Center Automation Wave Represents a Holy Grail Bigger Than Consolidation and Business Continuity Combined

Today the proliferation of x86 servers, as well as VMs, has created management chaos.  It has left some administrators wishing the days of the mainframe would return.  As an industry, we’ve collectively traded the centralized administration of the mainframe for the better price performance and greater departmental control over resources of client-server systems.  VMware, Microsoft, and Citrix want to deliver the best of both worlds.

This wave represents the biggest value creation layer thus far.  But it is not likely to yield the monopoly control and rents for anyone that Microsoft achieved with Windows.  Although it appears to deliver the control and switching costs of managing the underlying hardware infrastructure – servers, storage, and networking – it’s not clear it delivers the same switching costs for applications.

While this data center layer represents the emergence of a new operating system platform, significant differences exist.  Like traditional operating systems, the data center layer masks the complex and distinct elements of the underlying hardware infrastructure with software tied into interfaces it controls.  Like traditional operating systems, it provides services to applications that enhance their capabilities, such as availability – discussed above, scalability, security, and manageability.  Most significantly, it provides these capabilities to an application service that is actually made up of many components or tiers.  At the simplest level, it would ensure that the web, application, and database tiers all operate in synch to deliver the desired service levels.  Unlike traditional operating systems, however, it takes existing, unmodified applications and provides them with all these additional services.  In other words, the self-reinforcing networking effects coming from greater ISV support and resulting higher unit shipments don’t appear to exist at the application layer.

Figure 3:

VMware’s Virtual Data Center OS shares some of the properties of an OS.  It manages the interfaces to the underlying hardware infrastructure and provides services that enhance applications.  However, unlike an OS, it doesn’t require an application to write to its programming interfaces to leverage the application services.  Combined with heavy competiton from traditional vendors, that will probably limit its ability to achieve dominance via self-reinforcing network effects.  Source: VMWare

Pricing is not yet clear, but Microsoft won’t be giving this functionality away.  Unlike with their server virtualization layer, they have great revenue ambitions for their System Center management tools.  For VMware, the price point is likely to be higher than their current mid-range $3K / socket.   Unlike VMware, Microsoft has the opportunity to leverage the millions of developers using Visual Studio .NET.  As part of their Dynamic Systems Initiative, their management tools will be able to capture the information coming out of the developer tools that explain how all the elements of a distributed application work together.  Combined with a map of a customer’s infrastructure, they have a lot of the information they need to keep things running at customer-defined service levels.

Managing workloads at the Windows or Linux layer as VMware does, or the .NET layer as Microsoft does, isn’t the only approach to management automation.  IBM is taking J2EE applications and isolating and managing how they deliver on service levels at the application server level.  And the management software Big Four, CA, HP, IBM, and BMC are leveraging their technical assets in managing services, independent of whether they’re virtualized.  But these different approaches are a topic for a future post.

Desktop Virtualization Has the Potential to Remake How Users Interact With Their Desktops and the Business Model for Vendors

Conventional wisdom is that desktop virtualization is a fancier, richer version of what Citrix has delivered for many years with its Presentation Server product, now called XenApp.  It’s actually growing into something much different.  For starters, VMware, Microsoft, and Citrix are all trying to change how users interact with their desktops.  Today users’ desktop environments – all their settings, applications, and data – are tied to a physical box.  In the future, that environment will be associated with a user, following them around to whatever desktop or other device they might log into.  Citrix would appear to have the pole position in this market.  They can start by upgrading their existing 100m Presentation Server users and leverage the mature channel they’ve built to reach them.  In addition, they have Microsoft referring the desktop business their way.

But VMware is trying to go a big step beyond this approach to a place where Microsoft and their partner Citrix will be hesitant to follow.  The way Microsoft and Citrix are pursuing the desktop opportunity appears to assume Windows already resides on whatever desktop, laptop, or thin client a user logs into.  If online, the user’s environment is downloaded from a centrally managed set of servers.  If offline, the environment updates the centrally managed servers, to the extent policy allows, the next time the user is online.  This respects the very heart of Microsoft’s strategy, namely to put a copy of Windows on every box that ships anywhere in the world.  Beyond the basic Windows licensing cost, Microsoft currently charges $100 per year per client that remotely connects to a server.

VMware wants to break that business model by offering users greater choice.  In this approach, an OEM version of Windows doesn’t have to be fused to the client device first.  Instead, a bare metal hypervisor can host any version Windows, Mac OS X, or just a browser for running rich internet applications.  The idea is that users, not IT, can choose whatever client device they want, and IT just provisions a virtual machine that runs on the client with appropriate access permissions and applications.  Where Windows traditionally managed all access to the hardware and, therefore, had to be purchased with the box, now it’s just one of several potential personalities.  This doesn’t mean Windows goes away.  It does mean though that customers will have more flexibility when deciding about Windows upgrades or whether to accept Windows as the OEM-installed default OS.  Citrix could clearly do a bare metal hypervisor also, but would it venture into territory that would annoy its key partner?

What’s the business problem that a potential alliance needs to address?

1) The most obvious one is that current VMware Distributed Resource Scheduling (DRS), Disaster Recovery (DR), and High Availability (HA) functionality built on core capabilities like VMotion are incomplete. It’s hard to move the Virtual Machine (VM) for spare capacity or to deal with downtime to any random server and maintain the connections to the same isolated data and storage area network (SAN). Instead, administrators either have to open up the network so any server can see any other server and any storage device, a security risk, or they have to manually remap the connections.

2) The less obvious and more speculative problem to be addressed is the management and automation of business services across resources and applications. It is still primitive, though the big 4, CA, BMC, HP Openview, and IBM Tivoli are all hard at work addressing this, and CIOs are looking for the provider of a strategic, new provider.

So what might VMware and Cisco do about these challenges?

1) As far as DRS, DR, HA – Cisco could enhance its Nexus 5000 switch to make it possible for VMware DR, HA software to move a collection of servers, related storage, and associated network connections to a backup site or to add additional capacity with no manual intervention. Adding the mobility to virtual connectivity would probably take some collaborative development between VMware and Cisco, building on Cisco’s existing technology. The Nexus 5000 already virtualizes the I/O between a server and the data and SAN networks. It takes one big Ethernet pipe from the server (2 for redundancy) and then delivers separate connectivity to data and SAN networks.  (Starts 3Leaf and Xsigo offer this functionality and supposedly Brocade is close behind).

Some have suggested that Cisco could go so far as adding VMware VMs to server blades in its switch. we think that’s unlikely because it would create instant enemies out of HP, IBM, and Dell and motivate them to elevate competitive offerings. We think a more plausible alternative is to tip toe up to the point of adding full server capabilities.

Given that enterprises are spending more on server I/O and storage than virtualized servers,Cisco may see a larger market opportunity as well as a collection of less threatening competitors. Instead of invading traditional server territory, Cisco could potentially carve out much of the value add of the storage systems as possible to build on its dominance of connectivity.. Today, VMware DRS, DR and HA software works by orchestrating a workflow that involves software on the virtualized servers as well as software built into the storage system such as thin provisioning, snapshots, deduplication, and replication. It’s possible Cisco could build enough storage intelligence into its switch so that VMware can have it manage the storage intelligence such as snapshots and replication. In that scenario, the Cisco switch would treat the back-end storage as a dumb set of disk arrays and make multi-vendor storage systems much easier to manage.

2) On the management and automation front, VMware has big ambitions but little to show for them thus far. Cisco might be able to help there. VMware’s goal is to deliver the automation, reliability, and security of the mainframe to a data center full of commodity x86 servers. Cisco has a management and automation product called VFrame that has good functionality on paper but has experienced teething problems within Cisco. From what we’ve heard, Cisco’s own IT operation doesn’t run it and the sales force never warmed up to it. They appear to be more comfortable with selling big ticket infrastructure based on speeds and feeds instead of a business solution. Perhaps VMware can build on it and breathe new life into it.

Challenges ahead

Whatever happens, VMware has its work cut out for it. Microsoft appears to be aligning its significant systems management resources behind its hypervisor foundation.  Citrix has an enviable channel and installed base from which to upsell desktop virtualization. And even Red Hat has acquired a well-regarded, high-performance hypervisor. Meanwhile, the big 4 systems management vendors have been working on the management and automation of business services even while the virtualization revolution was brewing.

The Ultimate Objective

· It’s more than just the savings from server consolidation and more than just greater flexibility in managing planned (VMotion) and unplanned downtime (disaster recovery, high availability)

· Ultimately, it’s about automating the data center in order to make it easier for companies to deliver online business and consumer services. The iconic example of an online service that complemented a traditional business was the Sabre travel reservation system born in the ‘60s. It was based on purpose-built infrastructure that required intense collaboration between the customer, American Airlines, and the vendor, IBM. More recent examples include Fedex package tracking or the familiar dot.com services from Amazon, eBay, and Google. In order to make it easier for businesses to build or assemble end to end services from existing assets, technology vendors have to convert “assets” into “pools of services” using virtualization at every layer of the IT stack.

Looking at the IT Stack Layer by Layer

Servers: Steady Workload Expansion Suggests Near-Term Pricing Stability

· Continuing consolidation of virtual workloads has created servers with more and more multi-core CPUs. Whereas we’ve all heard of customers consolidating 10 workloads on a single server, we’re starting to see the intent to put 50 on a single server. Although customers want to stay within the envelope of commodity x86 servers, this steady expansion of individual server capacity suggests prices and margins shouldn’t completely evaporate. (Suppliers to public cloud vendors such as Microsoft or Yahoo or Google may find otherwise). But with this capacity expansion trend comes new risks and bottlenecks. For one, each server becomes too big to risk failure.

Storage: Accelerating Migration From Server Direct Attached Storage (DAS), Consolidation of Vendor Investments During Systems Refresh, Increasing Value of Storage System Software

· Storage should see a bump in growth from the convergence of the above factors though ultimately “thin provisioning” of capacity on-demand to applications and data de-duplication should return growth rates to historical norms

· With servers too big to risk failure, storage continues to be carved out and put on a NAS or SAN so that virtual machines can be easily migrated in the case of planned or unplanned server downtime while still pointing to persistent storage on the network. Apparently, customers used to frequently have SAN or NAS environments segregated by application. These are undergoing consolidation so that all the workloads or applications consolidating on fewer servers can all connect to one or a few pools of data.

· Customers also appear to be consolidating vendors to get closer to a single virtual pool of storage during this simultaneous refresh of the storage systems.

· Increased spend on backup, disaster recovery, and high availability is causing additional value to shift to the storage systems for several reasons. First, the number of workloads or applications considered mission critical or just important enough to ensure quick recovery continues to grow as it becomes easier and cheaper to deliver these capabilities. Second, backup and recovery capabilities have moved off servers. With so many workloads consolidating on them, this functionality is bottlenecking their operation. By living on the storage network or the storage system itself, they operate where the data already lives without soaking up server CPU cycles. Snapshots that get backed up off the storage system enable continual data protection. Replication to a storage system at another site provides the foundation for high availability. Data deduplication, however, prevents this profusion of data from growing geometrically.

Networks: Speed and Redundancy Driving a Refresh

· With the extra load on the path between consolidated servers and each other and more networked storage, the pipes are getting clogged. As a result, both appear to be in line for an upgrade. The data networks are likely to move to 10GbE. And the storage networks are likely to move to higher speed Fiber Channel over Ethernet or iSCSI.

· The big unanswered question is whether customers will consolidate their data and storage network equipment vendors so that it will be easier to configure and manage end to end secure quality of service (QoS) bandwidth from spindle to server to client.

Operating Systems: Just Enough OS (JeOS) at Just Enough of a Price

· In a virtual world, the server operating system becomes just a library of functionality that is part of a bigger container of functionality. In other words, ISVs can deliver a ready to run virtual machine (VM) file that contains all the operating system, middleware, application, and specific configuration settings necessary to run the application by deploying the VM file as a virtual server.

· In this “appliance” scenario, the ISV distributes or determines the configuration settings for the supporting software so that only the minimum required capabilities are used. Consequently, deployment of and pricing for any extraneous functionality is squeezed out. Today this works with Linux, but Microsoft prohibits it for Windows. We’ll see if they can maintain that position.

Server Software: Licensing More for Variable Usage

· Until a few weeks ago all major vendors such as Oracle and Microsoft licensed their server software for “peak” usage. In other words, customers paid for the maximum number of cores, CPUs, or servers their software would physically touch or run on. In a virtualized data center, however, capacity fluctuates more with demand and certainly more fluidly for maintenance and high availability activities.

· In this more fluid environment, there is intense customer pressure to change the licensing model. Just in the last few weeks Microsoft eased the licensing restrictions on its server software to accommodate deployment to virtual servers. This is likely the first step for all vendors to an environment where software is licensed more according to its usage than according to the number of servers it is deployed on. That would appear to be an effective price cut for customers.

Enterprise Applications: Lower TCO

· Virtualization doesn’t seem likely to up-end the economics of enterprise applications. But increasingly sophisticated management tools, starting with deployment, availability, and recovery, should help drive total cost of ownership (TCO) down to less burdensome levels.

Management and Automation: The New Data Center OS

· One major new layer in the IT stack will emerge over the next few years. Today we call it systems management, but it will go beyond the monitoring and management of traditional physical resource pools such as storage, servers, and networks. Instead, we are likely to see a management platform that automates and orchestrates the delivery of resources from virtual pools of infrastructure to deliver business application services according to policies set in SLAs.

· This layer is what Microsoft, IBM, and HP among others called utility computing early in the decade and what the public Web companies call cloud computing today. For it to work, however, it appears that existing software will have to be modified to be more deeply aware of the connection between how it’s designed and how it’s deployed. Like all major platform shifts, this will take time but create great value for its owner. Microsoft and VMware are both focused on this layer but maybe we’ll see a dark horse such as Cisco emerge.

We are very skeptical that VARs will be able to thrive and prosper in their current business model as SaaS adoption continues to gather momentum. The reason is that the TCO model that is promised by SaaS drastically reduces the revenue pool accessible to channel partners. (As we point out in our earlier post, the TCO model still needs to be proven in the long run.) In a June 2007 article, the McKinsey Quarterly compared the total cost of ownership (TCO) for a 200-seat CRM license as on-premise ($2.3m) vs. SaaS ($1.6m). More interesting though than the headline 28% reduction in TCO is the fact that the non-software revenue pool accessible to channel partners shrinks by 90%. Specifically, in this midmarket example, the $1.1m that is spend in the on-premise model for implementation, deployment and ongoing operations shrinks to a meager $106k in the SaaS world.

Faced with that bleak prospect, VAR channels can pursue one of three strategies:

Provide managed services. Become an MSP by offering management of the on-site IT infrastructure. Then shift as much as possible to offsite delivery. This is probably the most achievable and sustainable of the 3 strategies for a majority of VARs. In the infrastructure arena, this may be a way to escape the commoditization of storage and computing that drives margins relentless lower. The Holy Grail here is to price in a way that compares favorably with the high TCO of in-house operations. Let’s illustrate this with a numerical example from the world of data and systems management. Suppose you are a VAR that currently resells online backup for servers. Odds are the VAR’s customer may pay $0.50 per GB, so that’s $100 per month on 200 GB that are backed up from 1 server. Now suppose the VAR transitions to a managed services model by selling server management (i.e., provisioning, administration, system monitoring, maintenance, troubleshooting), including on-line backup. How would you price that? Let’s see. So one IT administrator (at $100k/year fully loaded cost) manages 20 servers, so that’s $5k/server/year of labor just for systems management (i.e., not including online backup). Assume an average server price of $4,000-5,000 with a 3 year life span, so that’s $1.5k/year, for a total cost of ownership of $6.5k per year per server, mostly in labor. Thus, the customer’s server TCO is ~$540/month. The MSP can promise a 30% TCO savings by selling a managed service at $380 per month, easily a much stickier business compared to just reselling products. Barriers to entry are pretty low though and are further reduced by emerging solution providers that cater specifically to MSPs, e.g., Digisense (“secure data management”) and Level Platforms (“managed services software”). While there are already an estimated 10,000 MSPs in the United States today, customer retention rates in this business are pretty high.

Extend the solution. Turn the one-size-fits-all SaaS offering into a customized solution by developing re-usable additional functionality related to vertical or niche horizontal requirements or integration with third party applications. A pre-requisite for this is a SaaS platform and marketplace provided by the vendor. This has limited applicability for infrastructure applications, while in the business apps space Salesforce is the leading example with Force and Appexchange. The platform ensures that VAR-developed solution extensions are compatible and reusable. The marketplace provides licensing and global distribution. And finally, the development environment lowers the barriers for the few channel partners with sufficientprogramming / scripting skills to pursue this strategy. Netsuite for instance encourages its VARs to pursue extensions. VAR deal sizes on Netsuite’s SaaS solution are $50-75k vs. $150-300 on a competing Great Plains on-premise solution. There is a belief that VARs/SIs who create vertical extensions can charge another $25K in high margin annually recurring subscription revenue, so not bad.

Ramp volume. Complement offline with online marketing and sales channels to reach new customers and increase share of wallet. Or acquire/partner with complementary VARs. This is the least sustainable of the 3 strategies given superior scale of vendors such as Dell or volume IT resellers such as Insight, CDW, or TigerDirect. Some larger and well managed infrastructure VARs will grow through acquisitions/partnerships and by extending their presence online. The volume strategy is debatable at best for infrastructure VARs (essentially amounts to running faster), and not really viable for business applications VARs given the consultative and industry-specific nature of the business.

SMBs have been the early SaaS suite adopters because traditional vendors couldn”t reach them

SMBs have been the low hanging fruit for early SaaS adoption because they’ve historically been underserved by application vendors.  Small deal sizes and bare bones cost of ownership requirements typically were critical stumbling blocks.  The small deal sizes mean vendors have to reach them with a much lower cost channel than direct sales.

So far there are two emerging channels.  The first is online, customer-initiated self-service complemented by inside sales that often becomes the viral “discover, learn, try, buy, recommend” process best articulated by Ray Ozzie in his 2005 Services Disruption Memo.  The second is further out because it requires a greater transformation among existing VARs.  Today they make a modest commission on the sale of traditional perpetual licenses or SaaS-based subscriptions. Configuration or customization and integration is where they make their margin.  It’s a time and materials business.  Expanding the market requires that in the future they deliver the same services to SMBs at a fraction of the cost.  The most likely way is to become mini ISVs by transforming their industry-specific expertise into an application hosted on and extending their SaaS provider, and use those higher margins to support much lower margin services.  That way the VAR should be able to squeeze down the subscription/service mix for their highly price-sensitive customers.

On the product side, SaaS flips the implementation risk from the customer to the vendor; the cost of ownership (TCO) experience is claimed by some to be as much as an order of magnitude lower.

Will suite emerging suite vendors such as Workday and Netsuite join SAP and Oracle in the enterprise?

Perhaps.  But there are still some obstacles in their path.

Why suites from at least some vendors are ultimately likely to emerge center stage in the enterprise

The TCO value of end to end process integration offered by suites is likely to emerge in the enterprise in the SaaS generation for the same reason best of breed collapsed in the client-server generation.  At first, the besat of breed products seem to offer better functionality, faster deployment, and more departmental autonomy.  But building and maintaining point to point integrations over time turns out to be a crushing expense.  For example, it’s not clear that CODA’s widely cited development of a Financials product on SalesForce.com’s platform means they can jointly deliver a reconfigurable order to cash process.  They appear more likely to share just the development and deployment platform, Force.com, not the end to end processes.  In the client-server generation, that would be the equivalent of running both on BEA Weblogic and assuming that means they’re integrated.  As in the client-server generation, however, the slow maturation of suites means the best of breed vendors will have a long time in the sun.

So what’s the hold-up with suites?

Why SaaS suites need time to prove their TCO advantage over best of breed SaaS applications

The favorable TCO economics for SaaS suites is nothing more than a hypothesis at this point based on the experience of the client-server generation.  Nor does that necessarily translate into an overwhelming advantage for SaaS suites over client-server suites.  After looking at the following numbers, you would think it would.  My friend Vinnie Mirchandani posted some provocative thoughts on traditional client-server TCO expenses, claiming it reached from $500 per user per month to as high as an astonishing $1,000 per user per month (cited in a presentation at Software 2007).  Quoting in full:

“Here are some numbers. If you spend $ 1 on a software license, over the next decade you will spend between $ 2 and 3 in annual maintenance contracts. You will spend 50c to $ 1 in project team training (vendor classroom training, travel costs, additional licensing of a training tool like RWD that SAP recommends), You will spend between $ 1 and $ 5 with the systems integrator and software vendor consultants (in some cases the services  may even be $ 10+ if you have a complex roll out). You will spend between $ 1 and $ 2 for every major upgrade – likely one every 3 years. So, even without factoring in costs of your own staff or incremental hardware you are looking at $7.50 at the low end and $ 21 (or more) at the high end for every $ 1 dollar of software license costs.”

For Business By Design, SAP’s new, purpose-built SaaS product, the company is claiming pricing in the range of $150 per user month.  But even if that’s a proxy for similar suite products such as Netsuite and Workday, that doesn’t mean the TCO is that low.  For example, just looking at Vinnie’s list above, there are still upfront costs such as project team and user training, business process reengineering, configuration, and integration and data migration from vendor and SI consultants.  Finally, nobody knows how difficult or expensive it will be to maintain the admittedly fewer interfaces to legacy or specialized systems across upgrades than it will be for best of breed SaaS applications.  Interfaces do change, even for SaaS suites, especially as the functionality evolves.  One issue no one disputes is that SaaS makes an enormous difference in the crushing cost of supporting older versions on an extended compatibility matrix of possible customer infrastructure configurations.  In client-server systems, that could chew up as much as 80% of the vendor’s R&D budget and untold amounts of customer operations expense.  With SaaS, that all goes into innovation that gets delivered to the customer as soon as it’s ready.

Hard evidence of the customer TCO advantage for SaaS suites in the enterprise won’t emerge for a period of at least another 5-7 years we would guess.  Even if the products were mature today, which is far from the truth for reasons we explain below, it would still take considerable time before they could fully penetrate IT operations at some major companies.  Only then, when they can measurably affect the IT cost structure of these customers over time compared to their peers running on different platforms, will all CIOs feel the pressure to meet the new, lower cost benchmarks.

Why SaaS suites need time to develop their go to market approach

Others have written about how for SMBs suite vendors can’t use the more viral adoption model aided by inside sales typical of departmental approaches.  While suites need to reach C-level decision makers, Netsuite, for example, claims that in order to scale more rapidly it needs VAR/SI’s who can make a business out of a first year subscription payment of $25K (vs $75K for Great Plains) and business process reengineering, training, and data migration services of $25-50K (vs $75-225K for Great Plains).  Relative to premise-based software, the service mix has to be lighter.  The only way they believe that model can work is if the VAR/SI’s can create reusable, vertical extensions that Netsuite hosts for which partners can charge another $25K in high margin subscriptions in order to subsidize a $50-$100K implementation project with minimal margins.

The challenge of selling direct with a richer services mix without viral adoption and while trying to build a channel has created entirely different economics for Netsuite vs. SalesForce.com.  Both are roughly 10 years old but Netsuite’s estimated revenues for 2008 are $160m vs SalesForce.com’s $1080m.  Netsuite lives on -0.05% operating margins compared to SalesForce.com’s 15%.  Both stagger under sales & marketing expenses that approximate 50% of revenue because of the slower reported revenue growth coming from subscriptions.  But Netsuite’s gross margin is roughly 10% lower than SalesForce.com’s, coming in at 71%, because of the richer service mix currently required to deploy suites.

Two other potential partner models include resellers who can reach a higher volume of customers in order to live on the lower margin commissions, or IT outsourcers who take over a customer’s entire IT shop, including SaaS applications, and become a managed service provider (MSP) for the whole infrastructure.

At the enterprise level, despite being able to charge customers for upfront commitments of one or two years given enough of a discount, the recognized revenue still gets pushed out, depressing growth and margins.  Bookings growth isn’t a substitute.  The only way to make this work is to gradually migrate revenues over to the new model over time.  Plus, vendors will have to overcome additional partner resistance.  The rich time & materials billings generated by custom implementations has to give way to a new generation of highly configurable software.

Why SaaS suites need time to mature as products

While this topic is worth a whole blog post on its own, several issues pop out.  First, there is the issue of the trade-offs between multi-tenancy and configurability.  Vendors such as SalesForce.com appear to tackle the problem of configurability and multi-tenant scalability purely in shared tables within one database schema instance.  They appear to have cracked the code on scaling out with commodity boxes rather than scaling up with expensive big iron while keeping the marginal cost of administration down.

It’s still unclear if they are disadvantaged in configurability.

SAP and Oracle, whether by virtue of having to drag along legacy technology or some other reason, appear to offer customers more isolation.  SAP gives each customer their own database instance and Oracle’s On Demand product appears to give each customer an entire application instance.  Keeping marginal administration costs low falls to the maturing management layer, the same type of technology that VMware claims will enable applications to meet service level agreements autonomously (e.g. without administrator intervention).  Like SalesForce.com on a different dimension, it’s still unclear if they are disadvantaged on the ultimate TCO front by virtue of high internal administration costs.  Giving each customer their own instance, however, has the added benefit of reassuring them that their data won’t be commingled with anyone else’s.

The other issue holding up suite maturity is simply the Mythical Man Month.  Suites take time to reach functional maturity no matter how many developers are working on them.  There have been reports in the blogosphere of complaints in the Netsuite customer base about product maturity 10 years after they started.  And SAP co-founder Hasso Plattner told a questioner at a Churchill Club event this spring that he would have to wait another 5 years for an enterprise-ready SaaS product from SAP, despite having started 4-5 years ago.

SaaS relevant for a class of processes today, but not all

For suites to reach both SMB’s and the enterprise, there needs to be both business model and product innovation.  Given the customer popularity and VC attention, many startups appear to believe that subscription pricing and hosting means they are SaaS vendors.  That just qualifies them as warmed over application service providers (ASPs).  This transition seems to be shaping up as a classic Innovator’s Dilemma.  Oracle says SaaS isn’t attractive because it’s a lower margin business that requires a new product and a new go to market model.  SAP seems to have reached the same conclusion but is plowing ahead gradually on both fronts.  While SaaS clearly appears to be working for departmental applications, the jury will be out on enterprise-class suites for several years at least.  Everyone agrees you can’t take an existing enterprise-class client-server suite and squeeze enough complexity out of it to reach SMB’s.  Nor, on the go to market side, can enterprise vendors immediately substitute a subscription pricing model for their traditional perpetual license one.  Netsuite seems unlikely to reach the enterprise by virtue of its design center and channel.  Meanwhile, Workday appears to be following the same steady strategy into the enterprise as its predecessor, PeopleSoft, without taking on SAP and Oracle in manufacturing this time.  In the end we believe the Innovator’s Solution in the enterprise is going to take some years more both on the product and go to market side.

But there’s one other thing that Microsoft doesn’t mention when it talks about purchasing just “search”: there really is synergy between search ads and display ads, which is where Yahoo gets the majority of its revenues.  It’s very hard to measure the value of display ads because they don’t surface when a user does something that shows their immediate intent (e.g. search for a lawnmower) – that’s why they can only charge for eyeballs/impressions while search engines can charge for clicks on keywords.

But everyone knows that those display ads create brand equity or other awareness that search ads ultimately get to monetize.  So all the major online vendors (MSFT, Yahoo, GOOG) who happen to have both major display ad networks (Google’s I think was built on Doubleclick) and search engines are planning to track what display ads users look at over time.  Then they will be able to correlate that with later search queries and search ad click-throughs.  In other words, they are finally going to start attributing more value to the display ads, which have never really been measurable.  But Microsoft has a plenty big display ad network so they just want to carve the search business from Yahoo.  And Yahoo is under such pressure to do a deal that they’re willing to sacrifice the future synergy of their search and display ad businesses.  Without search, they won’t be able to offer advertisers the same measurement capabilities as Microsoft and Google.  Their display ad pricing and revenues will suffer as a result.

They have been unable to articulate this synergy, or any other part of their strategy for that matter. It’s ironic, but they probably would be able to leverage that synergy with the Google search deal.  They also seem to be turning their internal platform for their properties into an Internet scale Web developer platform.  But they can’t even convince their top talent to give them some breathing room.  It’s game over.

• Serena Software is using Facebook as their corporate intranet and it now seems to be morphing into a sort of extranet. To overcome adoption challenges among its employee base, most of whom are ages 45 and over, Serena brought in a bunch of 16-year olds for Facebook Fridays. Serena’s SVP Marketing Rene Bonvanie claims 90% of employees are now using it. The primary benefit seems to be increased collaboration. Bonvanie says this makes it easy for both employees as well as customers to identify the right person for a specific question. Conversations have become more open and imbued with better knowledge. Thus, marketing and sales are losing some of their monopoly power as touch points with the outside world. In addition, knowing more about your previously face-less co-workers may also help increase a sense of common purpose at the workplace, states Bonvanie
• Best Buy’s Steve Bendt shared how their internally-focused ‘Blue Shirt Nation’ network helps generate recommendations that can increase store sales. Giving this online market place of ideas the look and feel of ESPN and online games was key to creating adoption among Best Buy’s young workforce, where turnover of 60% per year is the norm
• Paul Pedrazzi from Oracle shared how it’s internally-focused Oracle Connect and externally-focused Oracle Mix social networks do a great job of filtering content. One use case for Mix – once the traffic picks up more – is prioritizing customer needs and feature request prioritization. A common challenge is that product managers tend to overweigh feedback that is recent, local or comes from the largest customers – those who can afford to send their folks to Oracle’s executive briefing centers
• Shiv Singh from Avenue A / Razorfish shared how their own internal wiki, now used by 75% of employees, aims to increase internal information sharing. There is no silver bullet for overcoming employees’ tendency to ‘keep information close to their chests to impress their boss’. As you would expect, measures that can drive the right behavior range from executive sponsorship to making sharing fun to incorporating collaboration in the informal and formal reward and recognition systems

There was consensus that the 2 most commonly cited objections are not show stoppers in practice:

• How to deal with confidentiality and privacy? Relying on employee common sense and good intentions may be good enough for most situations. In general, people are smart enough today not to send out sensitive information over email or snail mail, not to leave reports on planes or the subway, not to dress in offending ways, not to burst into the CEO’s office for no good reason, etc. So with just a little bit of induction into the online world, why would you expect a different outcome? Best Buy cited that only 3 of 50,000 internal posts were offensive and had to be removed – pretty impressive given that their employee base tends to be rather young. Bonvanie says “97% of our information is destined for broad consumption, so open networks like facebook are a great distribution platform”. I would imagine that in time someone comes up for a compliance solution similar to what happened in email to ensure the remaining 3% does not leak accidentally
• How to make these solutions enterprise-ready? For internally hosted solutions such as Oracle Mix and Connect, ensuring security, availability, performance and compliance (e.g., with HR discrimination concerns) tends not to be a big issue. For externally hosted ones, that can be an issue. Serena’s Bonvanie again: “Facebook could pull any feature that we find really useful and there’s nothing we can do about that.” Personally, I predict we’ll see the emergence of white label enterprise class alternatives soon, similar to what happened with instant messaging (not that I know many folks who use those compliant solutions)

Could these 4 rather simple, early examples of ‘web 2.0 gone enterprise’ point the way towards more open, collaborative, connected value chains?

PS – I suspect the video from the session will shortly be available here….

http://www.uberpulse.com/

and the audio here…

http://churchillclub.org/podcasts.jsp