The IT triangle strategy for evaluating enterprise tech startups

There have been plenty of blog posts on why enterprise startups are “hot,” but enterprise startups can range from cloud infrastructure companies to developer software to new enterprise applications. Evaluating each market is very different because each of these markets are built upon each other (e.g., a next-generation CRM app is built using new database technology for analytics, which is in turn hosted on software-defined data center technology).

As a partner at Greylock, I’ve evaluated startups at every level of this enterprise software “stack,” and I’ve developed some frameworks for organizing how I think through the different opportunities. Two key ones for enterprise IT that I want to share are the infrastructure triangle and the enterprise applications triangle. I use the triangle to organize thoughts on how each technology is currently being disrupted.

The infrastructure triangle


Every computer, data center and even smartphone breaks down into the three main infrastructure components: compute, storage and networking. Overlaying all three corners corners is management and security. Each corner is facing disruption, and the axis between each area is where a lot of the action is. For example, the axes between storage and CPU, and between networking and CPU, is where I place software-defined storage and networking companies, because increases in compute power enable commodity servers to replace legacy storage systems from EMC and Cisco.

But let’s focus on each corner individually, starting with CPU. In one way or another, the IT revolution over the past 20 years has been driven by Moore’s Law, the constant improvement in CPU power and efficiency. Typically, cloud platforms and applications use virtualization to carve up a server into multiple virtual machines because applications have historically been written with the server or VM as the unit of compute. However, we are seeing the start of a revolution of how applications are built to take advantage of the improvements in CPU power.

The next generation of SaaS and scale-out web companies are writing their applications not to require VMs, but rather to use process- or container-based isolation to increase performance and density on their servers. As a result, we are seeing a resurgence of interest in lightweight container companies like Docker and CoreOS. These technologies sacrifice the strong isolation of a virtual machine in favor of higher density and performance; perfect for the scale-out generation of applications.

In addition, we are seeing open source projects like Mesos, which came out of the Berkeley AMPLab project, and YARN, which came out of the Hadoop ecosystem, to build and manage these distributed applications. Inevitably, the future data center will be a combination of all of these technologies, but the shift from VMs to a scale-out architecture has huge potential.

The biggest capital expense in almost any data center is storage. Storage is the underlying system of record for all our data in the cloud. We want our storage “deep and cheap.” At the same time, we want it fast.

To disrupt the storage corner, a new product has to be either 10 times cheaper or 10 times faster than the existing storage. Case in point: The current trend toward all-flash storage is driven by this need for speed, but smart storage vendors are combining flash with software to create incredibly fast storage arrays that are also price competitive with the spinning rust arrays of the incumbents. In contrast, Data Domain helped drive down the cost of storage by dramatically reducing the amount of space a company’s data and archives required. The current trend of software-defined storage and even object storage technologies like Amazon S3 are really about replacing proprietary arrays with cheaper scale-out storage.

The networking corner has long been dominated by Cisco and its closed system of proprietary ASICs and software. However, we are seeing a delayering in the networking stack. Over time, the networking market will look like the evolution of the server market, moving from closed systems to open systems that are managed more or less like customers manage their servers.

Startups like Cumulus Networks and Arista are quickly eating into Cisco’s installed base, promising superior price and/or performance. Cumulus is using its Linux networking OS to enable customers to manage their switches at scale like they manage their servers at scale, reducing the hardware and operating cost at the same time. On top of these existing and new switches, customers are using software-defined networking to build complete, agile networks across the data center.

Finally, in security and management, the obvious trend is simply providing control for the rapid increase in new technologies. As more of the three corners of the triangle move to a hybrid cloud (Amazon, OpenStack, VMware) and multiple vendors (Cisco, Arista, Cumulus), management and security concerns will only increase.

The enterprise-app “buyer’s” triangle


With infrastructure investments, I look at the intersections between compute, storage and networking, but with enterprise applications I like to look at the world through the lens of the buyer/user. This triangle is useful for identifying not only trends, but using channels to identify the customers for each market. In the three corners of the triangle we have CIOs, CMOs/VPs of sales, and CFOs, each with their own needs and interests. Along each side of the triangle, we see the flow of purchase decisions between each department, referred to here as CRM systems, support and service (or what I like to call post-sales CRM) and enterprise “back office or, as some like to call it, the “backwaters” of enterprise IT.

The CIO has been the traditional buyer of enterprise software, but with the advent of SaaS and cloud computing, the CMO/VP of sales, or the line-of-business executive, is becoming the main buyer of applications. In fact, the individual employee is becoming a key decision maker. In just a few short years, we have moved from a BYOD (bring your own device) world to a BYOA (bring your own apps) world where employees can go to the web or app store and purchase any one of a million applications. The CMOs, or VP of Sales even, are increasingly buying more SaaS apps and tools driven by the always-increasing need for revenue generation and creation of new customer experiences.

A close look at the buyers in this triangle is reveals the opportunities. Along the CRM channel, the move to the cloud has driven the M&A and product roadmap for with acquisitions of businesses like ExactTarget, Radian6 and Buddy Media. Other companies such as MailChimp, Marketo and Constant Contact have all carved a niche in identifying new opportunities along this axis.

The post-sales/post-purchase customer experience has become increasingly important in a SaaS world, leading to a rise in post-sales CRM software — for example, increasing your support capabilities with ZenDesk. In a mobile world, the customer engagement changes dramatically; you want to reach the customer in the application, at exactly the right moment. This move to mobile has created a new generation of mobile engagement and analytics applications like Urban Airship, HelpShift and Flurry.

In the muddy channels of “back office” software, SaaS companies have found room to swim and win against incumbents like Oracle and SAP. While not as “sexy” as CRM software, there are billions of dollars being spent by companies in this market and a huge opportunity for companies to help rethink how software can change how a business is run. For example, IT business-management services are changing the job of the CIO into a manager of IT services, while tasks such as service management and human resources are increasingly moving to the cloud.

These triangles are far from perfect and definitely not complete, but they are a nice way for me to start to organize my own thoughts around infrastructure and apps. Other areas like developer services and application infrastructure (e.g., databases, PaaS, etc.) require different frameworks altogether. There’s more than $450 billion spent each year between data center infrastructure and enterprise applications, which means we need to draw some pretty big triangles.

Jerry Chen is a partner at Greylock Partners and former vice president at VMware.