How To Find a Customer in a Divided IT Universe

The last few years have seen a significant shift in the locus of innovation, which is having a meaningful impact on emerging vendors and startups. It used to be, if we wanted to talk to cutting edge users, we went to CTOs on Wall Street. Today, much of the most interesting technical innovation is taking place inside the largest scale web properties and cloud providers. The scale at which these companies are operating is forcing them to push the limits, rethink, and in many cases rebuild almost every layer of the technology stack.

This is creating a rift between the web and enterprise stack, which presents an interesting conundrum to startups. The web giants have the most interesting technical challenges to solve, but they tend to solve them through a combination of internally developed applications and tools plus open source products, loathing proprietary hardware. They may not make great customers. Enterprise companies will spend the money, but the sales process is long and expensive. The product definition generated by talking to the top 10 web sites likely won’t be the right fit for enterprise.

These changes in both the source of innovation and the purchasing habits of the large enterprises are having a massive impact on the entrepreneurs walking in our door. A business plan focused on selling exclusively to Google and Facebook is just as likely to fail as one focused on JPMorgan and Goldman Sachs. However, there are common threads that most technology organizations wrestle with, which span both big web and traditional enterprise. These include:

• making sense of massive data sets,
• figuring out how to integrate solid state drives into their infrastructure, and
• automating the provisioning and operations of thousand of servers and network elements.

In these segments and others, we see great opportunity in unlocking and productizing the lessons learned in scaling the web and making those lessons more broadly available.

Scale and speed are the mantras of this new breed of Internet performance engineer and developer. There are whole conferences devoted to scaling, including Structure and Velocity. Pager-wearing web operations folk are the new celebrities in Silicon Valley. Web 1.0 was built on the 4 four horsemen: Sun+Cisco+Oracle+EMC.  Today, that’s anncient history. There’s now a massive dichotomy between the open-source/internally developed software and commodity hardware stack that runs the web and the large ISV plus Dell/HP/Oracle/Cisco/VMware/EMC/NetApp stacks the runs the Fortune 500.

The recent economic downturn has only pushed the Fortune 500 deeper into the arms of the technology giants, Oracle, Cisco, HP, EMC. Large enterprise has been willing to spend on high gross margin products and pre-integrated solutions if it means they can reduce their operational headcount. The web’s start-up DNA drives frugality and pushes web startups in the opposite direction, spend as little as possible on capital expenditures, ride the commodity cost curves, and hire smart people to glue together the cheap hardware with brilliant software.

Of course, the web giants are not just consumers and integrators of technology, but increasingly developers, too. This is due to both addressing unique sets of scaling issues, and to deriving competitive advantage. Frequently, these companies are also contributing innovation back to into the open source community. The myriad NoSQL projects are just one example of this trend. Many take things much further: modified Linux kernels, custom compilers, internally developed networking control planes, even building their own servers and switches.

So how does a start-up bridge this stack gap? We see a number of different approaches:

• Ignoring the gap. Focus on something horizontal enough to apply to enterprise and web, such as the emerging memory and SSD tier, big data analystics, etc.
• Twists on open-source. Use open source to get distribution and evolve a business model beyond service. Opscode and Drupal are two interesting examples.
• Trickle down. This approaches aims to “productize scale.” Take the lessons learned scaling the web and apply them down-market into the enterprise as they become more web/cloud like.