Google’s Chrome OS announcement has lit up the blogosphere; the OS is expected to hit the market late in 2010. Most of the excitement seems to center on the fact that this is the first genuine salvo in the persistently-rumored titanic showdown between Google and Microsoft. As much as the pundits’ would like to see some serious competition for Windows–a desire I fully support if only to see what Google’s innovative take on an OS might be–it is worth looking over Google’s own product history and some key factors that have led to Microsoft’s dominance of the desktop to divine whether Chrome OS stands a chance.
Microsoft’s dominance of enterprise and consumer desktops has come about because of their OEM relationships and their immensely successful Office productivity suite. Microsoft has one of the widest set of OEM partners in the industry and it’s certainly not going to look on benignly if Google attempts to capture some of those for itself. Let’s remember that this is the same Microsoft that dominated and in some cases coerced major OEMs to play primarily by its rules in what might in principle be a co-equal relationship. Google, relatively speaking, appears to be a loner that is used to agenda-setting roles in its existing relationships. Google faces an uphill task not only in cultivating this OEM relationship-building DNA but also to emerge as a serious alternative that OEMs can offer enterprises (let’s not forget the enterprise is where Microsoft makes most of its money).
Which brings me directly to the second reason Microsoft has been successful: because enterprises actually want Office. Billions of dollars worth of enterprise value are riding on the strength of Office productivity applications. Although Google Apps introduces a fundamentally new angle to collaboration in productivity applications, I have not come across too many enterprises that have switched entirely away from Office (or Office clones) to Google Apps. This hesitation is not just because Google Apps don’t have all the productivity features enterprises expect but also because the impedance mismatch between desktop and Web applications has not yet been crossed compellingly. Chrome OS has to be a lot more than a conduit to deliver Web-based browser applications – it must also integrate with cloud storage, desktop search and task management. Until Google can make a compelling case that the user experience of today’s desktop/laptop-bound knowledge worker won’t be altered for the worse, it will have trouble pushing product.
Add to the above the fact that Google often seems out of its depth when selling packaged products to consumers or enterprises. Not only is the concept foreign to the company’s hip, Webby DNA, a simple Google search will show that the company has been found lacking in responsive customer service even for its paying customers. Google may well give away the OS to OEMs as a loss leader to complement its Web portfolio but supporting it nonetheless requires focus and staying power from a company that has left several initially promising products languishing by the wayside.
When Chrome was initially announced, I was intrigued by its inventive approach to Web-based computing. I’ll continue to watch Google with a cautiously optimistic eye because it’s one of the few companies that realistically stands to reinvent the desktop OS experience. A lot of things may appear not to be lined up in Google’s favor at this point, but a lot of things can change in the year and a half that’s yet to pass before Chrome OS comes out.
Cloud computing and startups
Kudos to Intuit for hosting a great event “Startups and the Cloud” at Bentley College on June 11. I was invited to be a keynote speaker at the event and I had the opportunity to lay out the cloud computing landscape, especially highlighting what it meant for startups.
Upon request, I posted a slightly modified version of the presentation I used at my talk to Slideshare, and it’s embedded below. Video of the talk is said to be forthcoming; I’ll update this post with that information when I have it.
Are enterprise clouds the new SOA?
Have you heard of this cool new concept in enterprise computing that will revolutionize the way your organization’s IT works? It will meet the technological needs of your business of course, but in addition to that, it will save you tons of money. By pooling and refactoring your infrastructure resources, it will save you the busy work of building out your computing infrastructure and let you concentrate on what you do best in your business—adding value. If your eyes are glazing over, I’ll stop right there. Maybe you think I am talking about cloud computing, but practically the same arguments were made in support of service-oriented architectures half a dozen years ago.
I should reveal my cards right away in that I don’t think cloud computing in the enterprise will be an albatross like SOA. Cloud computing is not encumbered by many of the same issues that hamstrung SOA rollouts in the enterprise. Moreover, the incentives seem to be stronger because the technical and business benefits of cloud computing appear to have been articulated more clearly. I am getting ahead of myself though. As a member of the tech investor and industry analyst circles over the past year, seeing the innovation ecosystem emerge around cloud computing awakens an uncomfortable series of flashbacks in me.
For one, the excessive emphasis on an elaborate taxonomy for what is essentially an ambitious, yet nebulous concept. SOA’s technical promise did sound alluring, but the effort of building a top down, exhaustive taxonomy quickly overtook the effort of meaningfully proving these ideas out in the field. By the time it was apparent that enterprise SOA efforts were lagging behind, way too much intellectual and emotional effort had been invested into taxonomizing to throw it all away. Instead, we were subjected to efforts like the WS-* standards and a bevy of startups in frightfully tiny niches like SOA governance when real world implementations were seriously lagging behind. Those efforts weren’t entirely in vain given how they have found their way into some SOA-related products from large software vendors; however, at least some of the pure plays that mushroomed around SOA had their origins in the extensive taxonomy effort. What does this mean for cloud computing startups looking to catch the eye of technology investors? Rather than appearing as a me-too startup in one corner of a vast taxonomy with a solution looking a problem, create your own market segment where your customers can speak powerfully to the value you add to their business.
Another flashback to SOA comes from the oft-cited cultural change that must occur if cloud computing is to succeed in the enterprise. In addition to the significant technical refactoring that is part of any SOA rollout, SOA advocates emphasized the “human side of service-orientation”, i.e. redrawn spheres of influence and ways of working that an SOA implementation would also imply. To date, enterprise efforts within the cloud computing phenomenon have largely been one-offs or short term cloudbursting implementations driven largely by small, technology-savvy teams. Sustainably running large portions of a business in the cloud, however, requires new ways of looking at resource consumption and allocation, both of which require significant involvement from accounting, finance, IT and several other key business stakeholders. Startups pitching a cloud management message to enterprise customers and VCs: keep in mind that customers will appreciate some indication of the requisite cultural and organizational changes in addition to your technology and product differentiators as they perform their cost-benefit analysis.
The cloud computing phenomenon shows the classic signs of an early market. Most enterprises view a cloud computing provider as little more than a managed hosting hosting provider with a less restrictive contract. But we believe the promise of the cloud to be much more than just a new delivery mechanism for enterprise computing. The startups that will be most successful in this regard will be those that bring about fundamentally new kinds of applications and services by leveraging the cloud. Perhaps a new application development model that obviates the black magic that goes into application deployment? Or perhaps a new kinds of analytical applications enabled by the sea of datasets available on EC2? We don’t claim to have all the answers but if you are an entrepreneur who does, we’d love to speak with you.
Our investment in VKernel
Today, VKernel announced a new round of funding led by Longworth Venture Partners. We welcome CEO Alex Bakman and the VKernel team into the Longworth family. The press release and other relevant details can be found here.
We believe virtualization represents a major platform shift, similar to prior shifts to client-server computing, Web applications and n-tier enterprise applications. Each successful platform shift anoints its own winners, including the maker of the platform. Users of new platforms invariably need tools to manage them, and management tools for successful platforms can themselves represent compelling market opportunities. Platform makers, however, don’t always end up owning the market for managing their platforms because management concerns are often orthogonal to platform concerns. Historically, every classically disruptive platform shift has produced its own flourishing crop of systems management companies. We’re betting that VKernel will be among the systems management hits birthed by the move to virtualization.
We have been impressed by the traction VKernel has found among administrators of virtualized server environments. VKernel eases the process of capacity planning, change modeling and chargeback in virtualized environments by providing busy administrators with a suite of usable, high-quality management tools. Each tool is an easy-to-install virtual appliance that provides valuable insights in hours. VKernel has received enthusiastic recommendations from its customers and was recently named to the InformationWeek Startup 50 and as one of CIO.com’s 10 Virtualization Vendors to Watch in 2009. Having started and grown systems management companies like Ecora Software before, Alex has a strong understanding of the market as well as what it takes to scale a management tools business. We look forward to working with Alex and his team as VKernel brings out more virtualization management tools to fill out their vision.
Virtualization is an active area of interest for Longworth, and VKernel joins high availability startup Marathon Technologies in our portfolio. Past investments of Longworth include application virtualization startup Softricity, now known as Microsoft App-V.
BarCamp Boston
I’ll be attending BarCamp Boston at MIT’s Stata Center on April 25-26, 2009. Go to the BarCamp Boston site and register for a weekend of ideas, cool sessions and general geekery. There seems to be a pretty good mix of topics spanning a wide spectrum of interests. From the website:
Topics may include, but are not limited to: open source software, startups, UI design, entrepreneurship, AJAX, hardware hacking, robotics, mobile computing, bioinformatics, RSS, Social Software, programming languages, and the future of technology.
I plan to be there for at least some part of both days. Come find me and say hello if you’d like!
Thoughts on the Open Cloud Manifesto
Late last week, I got hold of a leaked copy of the Open Cloud Manifesto, a proclamation and call-to-action for all stakeholders in cloud computing to keep in mind the principles of openness and interoperability when advancing the state of the art in cloud computing. The sentiment expressed in the document is admirable and no doubt one in which we have a deep vested interest as technology investors. I am also attending the Cloud Computing Interoperability Forum (CCIF) event this week. It is only germane then for me to share some of my thoughts on the document.
For a document coming out in 2009, I must admit I was a bit underwhelmed by its content and scope after having been barraged by the term ‘cloud computing’ in the past year. On the other hand, the time has certainly come to lay down a foundational document which ties together concerns paramount in stakeholders’ minds and forms an operational baseline for future conversations. I’m really happy that the document does not get bogged down in semantic games trying to define cloud computing, an almost intentionally vague term whose definition, much like ‘distributed computing’ or ‘artificial intelligence’ of yore, is best expressed as the emergent effect of a number of enabling technologies. By laying out the value propositions of cloud computing in the definition section, the document has chosen to take a pragmatic approach to defining the term.
Of the four value propositions of cloud advanced in the document, we are naturally most excited about the first and last ones as investors. Aside from the lower startup costs and greater capital efficiency of “cloud-native” startups, being able to scale on demand means higher availability for customers and lower fixed costs for investors. Moreover, these two value propositions are entirely new and disruptive to enterprise computing, thus raising the possibility of new business models to fill the resulting innovation gap. I am skeptical of the other two value propositions: streamlining the data center and improving business processes because they are goals enterprise computing has been working towards for decades now. Although as investors we’re naturally biased towards disruptions rather than incremental improvements, even sustaining value propositions open up immense market opportunities for the creative entrepreneur.
Buried deep in the document is perhaps its most important sentence: “As cloud providers ask their potential customers to accept a loss of control over their resources, hiding vendor lock-in behind the benefits of cloud computing will lead to long-term damage in the cloud computing industry.” In addition to boundaries of resource control being redrawn, several technological changes, such as post-relational databases, are happening at the data and application layer, which will bring the issue of vendor lock-in to the fore. In such transformative times for enterprise computing, the manifesto must take care not to conflate openness with standards-compliance. Once the initial cognitive barrier to adopting cloud computing is overcome, the benefits of cloud computing are so immediate that a de jure standard will probably not stand in its way.
As with any technology-heavy solution within enterprise computing, it’s just as important, if not more so, to pay attention to what works rather than mandate standards compliance in a top-down manner. The document is important because it is one of the first to start a general conversation about an open cloud. What’s more, it has already taken a pragmatic approach to defining cloud computing. Its authors would do well to continue in a similar vein and keep their ears close to the ground to learn what works de facto rather than get discredited and lose their initiative because of pious insistence on de jure standards compliance.
If 2008 weren’t enough already, buzz around cloud computing startups is reaching deafening levels this year. On the one hand, large enterprises are taking the opportunity to reduce capital expenditure during the ongoing economic disruption. On the other hand, as abstract thinking around cloud computing crystallizes into concrete action, more and more startups leading with a cloud computing message are coming out of the woodwork.
Longworth follows a research-based approach to early-stage technology investing, where we form market maps and investment theses for particular market segments as we investigate them for startup opportunities. For most of last year, we put qualified investment opportunities with a ‘cloudy’ pitch in the same bucket as our those with a pitch around virtualization. There was a lot of messaging overlap last year between the two areas to be sure, because virtualization projects were being rolled out in a number of enterprises, and the ultimate goal of several of these projects was to build a private cloud. Fundamentally though, cloud computing, while being enabled by virtualization, is a separate area of investigation in our investment research (I had been doing some thinking about cloud computing at my previous gig).
In 2009, cloud is finally (and naturally) breaking out from virtualization and coming into its own as a distinct segment within our startup tracking processes. Because we’re actively investing from our recently-closed third fund, I’m going down the list of companies we used to track as virtualization startups and in some cases, reclassifying them as cloud computing startups. This reclassification has been an illuminating exercise in itself because I’ve had to articulate why I am moving startups from one list to another. Much like this (somewhat tongue-in-cheek) list of what can’t be cloud computing, here are some cues I used during this reclassification:
- Pitch. Does the company mention ‘virtualization’ at all in its primary value proposition, aka pitch? If yes, then its offering is virtualization-focused rather than cloud-focused, although it may be an enabler for an enterprise cloud. Note that I don’t automatically classify companies that mention ‘cloud’ as cloud computing companies because even a lot of virtualization-focused startups will use cloud to gussy up their pitches.
- Buyer Profile. Can a tech-savvy individual contributor within an organization at a departmental level ostensibly get started with the company’s offering, or does centralized IT need to get involved? The former would qualify as a startup with a cloud computing offering.
- Purchase Method. Does a customer have to undergo a traditional enterprise software purchase process, with purchase orders and heavy-touch, potentially lengthy sales engagements, or can they whip out a credit card and get set up in less than an hour? The latter would qualify as a startup with a cloud computing offering.
- Cloud Economics. Does the offering adhere to cloud economics, i.e. ‘pay-for-what-you-use’, even if it isn’t pay-as-you-go spot pricing? Does it include agile system provisioning as well as de-provisioning? If not, then the startup doesn’t qualify as a cloud computing vendor.
Not all of these cues cleanly demarcate the cloud startups from the virtualization startups. Sometimes I’ve had to use my own judgment, but the cues above do push me closer to an answer.
No doubt I’ll continue to discover other cues as I proceed down our list of startups. To entrepreneurs and others in the know reading this, are there any big ones I missed? If you’d like to suggest some, please join the conversation in the comments.
Amazon just announced a new feature in its EC2 cloud computing infrastructure: Reserved Instances. Reserved Instances enable Amazon Web Services users to reserve EC2 capacity with a one-time advance payment and pay less for ongoing usage, rather than relying entirely on the existing pay-as-you-go model, which the announcement calls a “spot market” for computing capacity.
This is great news for AWS customers for a couple of reasons. For one, they could save quite a bit on the cost of running an EC2 instance—one of the commentators at the original blog post above works this out to be 33% for a one-year commitment. Customers aren’t obligated to run a reserved EC2 instance full time and aside from the one time payment will pay for usage at a lower rate. What then does the one-time payment buy customers? A guarantee from Amazon that the launch of an EC2 reserved instance will succeed. When combined with EC2 availability zones, this gives enterprises a way of doing cloudbursted disaster recovery, which Amazon indeed cites as one of the motivations for EC2 Reserved Instances.
On a somewhat related note, there have been some interesting conversations lately about the effect of the fine-grained pricing models enabled by cloud computing infrastructure on application architecture. Pay-per-use models are breaking the cost impact of application architecture decisions down to levels of detail scarcely seen before in enterprise computing. It used to be that cost-benefit analyses of certain IT architecture decisions could be meaningfully done only in business-critical settings like financial services or telecom, and that too only at a coarse-grained level. With growing comfort around fine-grained costing and usage models (for a recent example, see Diomede Storage’s offering), we’re effectively looking at a future where the CIO and CFO functions in a large enterprise will converge.
All these developments are great for the innovation ecosystem around cloud computing, because it indicates that enterprises are getting more comfortable with a model of computing rather unlike anything that has come before, both on technological and economic grounds. The gold rush to build out cloud infrastructure may only have a few winners, but as enterprises move large swathes of business workloads to the cloud, the Levis of cloud computing are likely to see big business. We are excited about the possibilities this raises in areas like cloud management, offerings for SLA-based IT service providers and cloud costing, among others. Entrepreneurs, rev up your engines and get in touch if we can help you!
Virtual reality: Look at the market today
As we have seen via announcements from the major virtualization vendors over the last year, the virtualization platform is becoming a zero-ticket item, and the focus of innovation is moving to the management layer above the platform. We’ve recently been trying to wrap our heads around the virtualization management market, and what an illuminating exercise it has been!
Right before I joined Longworth, I was an enterprise IT industry analyst at The 451 Group, where virtualization was one of my coverage areas. To be sure, virtualization is turning out to have a truly transformative effect on enterprise infrastructure for a number of reasons. As an IT industry analyst, I was part of the class that was shaping the narrative that would define future conversations in enterprise computing. The heady conversations I was part of had to do with management issues never before seen on the x86 platform before the advent of virtualization. It was important, however, not to lose sight of the fact that virtualization is only just entering mainstream adoption, with an average of 10-20% of business workloads now virtualized.
During recent conversations with virtualization customers of varying sizes, we’ve come to realize that among several small and medium businesses virtualization of new servers has come to be the rule rather than the exception. This enthusiastic uptake of virtualization undoubtedly brings along its own set of problems; some of them are reincarnations of management problems from the physical world, whereas others are centered on new possibilities raised specifically by virtualization, such as resource pooling and enhanced automation.
Even just those areas of virtualization management that have counterparts in physical systems management constitute a large market today in dollar terms with several sizeable niches that would be lucrative to an entrepreneur. The existence of several closely-related niches like backup/HA/DR, capacity management, VM optimization etc. may make virtualization management look like a crowded market, but individual niches often don’t have more than a couple of vendors serving them. Moreover, this market exhibits classic signs of an early market, as evidenced by a puzzled user base that isn’t always able to articulate fine functional distinctions. Note that we aren’t even talking here about future management issues raised specifically by virtualization, which we also think will be substantial because of the transformational nature of virtualization. In fact, the IT analyst firm Gartner projects the overall server virtualization market size to reach $6.8bn by 2013.
So what does this mean for an entrepreneur looking to enter the virtualization management market today? Only that it’s hard to overstate the importance of that old business maxim: ‘listen to your customer’. Whereas it is important to be aware of conversations in the analyst community, which raise issues that will shape the future of your market, it is just as important, if not more, to listen to your customers and prospectives and see if there isn’t a problem you could be solving for them today both to their advantage and your profit. And if we could help you scale your business, we’d love to hear from you.
Virtualization and new business models
At Longworth, we take an active interest in virtualization, which is one of the hottest markets in systems software today. My role at Longworth is to stay on top of trends and interesting companies in systems software. I recently attended a talk by Mendel Rosenblum at MIT titled The Impact of Virtualization on Modern Computing Environments. I didn’t necessarily go there to learn anything new about virtualization, but I wanted to hear what the father of the dominant player in x86 virtualization had to say on the subject.
As expected, I didn’t learn anything new, but Prof. Rosenblum tersely and elegantly reasoned about virtualization before an audience of computer science academics. In addition to hearing him describe virtualization without industry jargon, I also came away with an interesting way to view ongoing shifts in enterprise software delivery through the lens of virtualization.
Prof. Rosenblum detailed five distinct roles in how software is consumed today:
- Computer Hardware Procurer
- Application Developer
- Application Configurator
- Application Maintainer
- Application User
When software is delivered as packaged software, one organization is responsible for all the above functions except the Application Developer function. This organization buys hardware, configures it, maintains an application and provides it to an internal user community. When software is delivered as a service—as in the SaaS model, by Google or Salesforce—one entity takes care of all but the Application User function by procuring hardware, developing software, configuring an application and maintaining it on an ongoing basis.
Now, Prof. Rosenblum said that these two are but two extremes of a spectrum. Other points in the space defined by these roles can be accessed via virtualization, because it decouples different layers of infrastructure. One example he advanced was Platform as a Service (PaaS), where Application Developer and Application User may be one organization whereas the grid on which the application runs belongs to another organization. Other combinations are also possible. He also raised the possibility, via the emergence of virtual appliances, that new roles may be created in the software delivery value chain, including Virtual Appliance Builder, Virtual Appliance Maintainer and Execution Platform Provider.
Prof. Rosenblum’s explanation struck me as a particularly elegant way of partitioning the space of software delivery. Additionally, I thought it was a particularly succinct way of explaining why we at Longworth are interested in virtualization. The decoupling of different layers of infrastructure creates new opportunities for product vendors and service providers where none used to exist before. For an example, see the above graphic from desktop virtualization vendor Desktone, which enables a 3rd party service provider to deliver desktops as a service to enterprise end users. Furthermore, these vendors and service providers can take advantage of new business models born from the technological changes that virtualization enables.
We are very interested in hearing about emerging business models in infrastructure enabled by virtualization, and we believe—based on our own market opportunity estimates—that what we see out there today is just scratching the surface. Virtualization has been around since the 1960s when computers (mainframes) were rare and expensive but viewed through today’s lens, it represents a significant platform shift for commodity hardware. We believe the best is yet to come from startups and independent vendors in terms of possibilities. If you’re an entrepreneur playing around with an idea in this broad segment, we’d love to hear from you. Leave a comment or drop me a line at vishy at longworth dot com if you want to bounce an idea off me.
Longworth Venture Partners Blog 