Music orientated so when hip-hop was originated
Fitted like pieces of puzzles, complicated
—Rakim, Microphone FiendArtist: Eric B. and Rakim
Track: Microphone Fiend
Album: Follow the Leader
As former founder and CEO of Loudcloud, one of the original cloud computing companies, I am considered somewhat of a domain expert in all things “cloud.” Combine that with cloud being one of the hottest areas of new development and the most overloaded term in technology (well, maybe “real time” is more overloaded, not sure) and I have probably seen more new cloud companies than any other variety. So, I thought that our first investment in the sector might be worth discussing.
As with previous major architectural shifts in computing infrastructure, when moving from traditional on-premise computing to cloud computing, certain things translate well, some don’t transfer at all, and most things are just different. When the computing architecture moved from Mainframe to Client/Server, hierarchical databases didn’t move with it. Due to radically cheaper computing cycles and storage, the database market moved to the much more easily programmable and more resource intensive relational database model. Similarly, the first attempts to build applications in the cloud from companies such as Corio simply fork-lifted the leading on premise software and moved it into the hosted environment. While this sounded like a good idea to many VCs at the time, it turned out to miss important details and advantages of the cloud.
The fact that these missing details mattered was definitively proven when Salesforce.com freshly architected their product for the cloud. This new sales force automation product boasted multi-tenancy, a native web interface, and an underlying operational environment designed for the modern era. Based on this design, Salesforce.com radically raised the bar for product and financial performance in the software industry. Their breakthrough architecture enabled them to run engineering at half the percentage of revenue of both their online and offline competitors. Salesforce.com took their winnings and invested in sales and marketing at levels never seen before in the industry. In the meanwhile, their multi-tenant approach enabled them to sidestep an issue that had dogged on-premise enterprise software for years—supporting multiple product versions. By hosting all of their customers on a single version, Salesforce avoided huge investments in multiple version support, migration, and upgrades then used those winnings to deliver exceptionally fast product innovation. By combining their innovation advantage with their aggressive sales and marketing efforts, Salesforce.com took the market and remains the most successful cloud application in the world 11 years later.
We recently made our first investment in the cloud in a cloud application company called Okta (which means cloud cover). Our investment thesis starts, as always, with the entrepreneur. The founding CEO, Todd McKinnon, previously ran Salesforce.com’s engineering organization from the early days (8 engineers) to large scale. He is the ultimate domain expert on building cloud applications. He may also be the most astute engineering manager that I have ever worked with. A mutual friend of ours, Hosain Rahman (CEO of Jawbone) refers to Todd as “the engineering management Ninja”). I must agree that Todd’s skill, like a Ninja’s, seems to give him some kind of super powers.
Okta’s initial product aims to solve the Cloud identity problem. I like the problem, because it is seriously different in the cloud than it is on-premise and, for new companies, different equals good. How is identity different in the cloud? For starters, problems that were solved long ago on premise such as automatically getting access to your mail, files, and printers remain issues in the cloud. Next, problems that were intractable in the on-premise world such as meta-directory—the technology that integrates disparate identity sources—become manageable in the cloud. Every system that a meta-directory integrates must be upgraded every year or two. When vendors put out these upgrades, things change. As a result of these changes, on-premise meta-directories break and must be re-integrated. For customers, this means paying for on-going, permanent integration consulting or punting on the meta-directory problem. Like a bunch of overly nervous football coaches, almost every customer in the on-premise world chooses to punt. In the cloud, the maker of a properly architected meta-directory can perform the upgrades on behalf of all of their customers simultaneously.
In order to understand the size of the market, it’s important to realize that in the on-premise world a variety of factors lead to an artificial separation between the identity market and the desktop and systems management markets.
The goal of an IT organization ought to be to optimize employee productivity. Nonetheless, every leading on-premise desktop and systems management product is designed to optimize machine productivity. Specifically, the primary objects in these systems are machines. People, on the other hand, are peripheral, loosely related attributes. This architecture appears to be upside down. In the on-premise world, it is necessarily upside down due to, among other things, the shaky foundation for people—the broken meta-directory.
In the cloud, you can solve the meta-directory problem and other related problems then unite the identity and systems management markets. As a result, I really like the market, because the cloud identity market will likely become the cloud management market. Given that those current markets combined are between $10B and $20B (depending on what you count), the resulting market will be extremely large.
However, this rough analysis understates the total size of the Okta opportunity, because Software as a Service (SaaS) will greatly expand the number of companies who can afford to deploy enterprise software applications. In the old world, large-scale enterprise software systems were extremely expensive to build and distribute. As a result, they were also very expensive to buy (small companies didn’t buy and deploy Siebel). With the advent of SaaS, smaller companies are adopting software applications in large numbers—the average Okta customer has less than 2,000 employees and already a few software systems that require integration.
We could not be more excited to be going after a giant market with the very best team in the field.