November 3, 2014
Ashu Garg
OpenStack, the popular open-source cloud orchestration platform, celebrated its fourth birthday in July and its tenth release, Juno, on October 16. This is a big milestone. The Juno release has taken the practical realities of enterprise IT, managing mission-critical applications, new software development, and shadow IT within security regulations and business constraints.
To get to this point has been quite a journey for OpenStack. What started out as a group of system administrators and NASA engineers sitting around a table talking about an open-source solution to replace Eucalyptus and VMWare — is now being used in production by companies like AT&T, Comcast and Sony. However, if you double click into the actual applications running on OpenStack today, they tend to be fairly simple workloads.
I’ve watched OpenStack grow up as a framework. What initially had too many high-level cooks, with integration, scalability, availability, and performance issues, is now about to turn the corner into a real enterprise grade framework for the first time. The recent jumbo financings of Mirantis and acquisitions of MetaCloud (Cisco) and Cloud Scaling (EMC) validate that enterprise IT vendors are taking it very seriously.
And even though its execution isn’t where it needs to be, enterprise customers really want it — or something that leverages it — to succeed. To understand what enterprises really need today, let us step back and take a look at the big picture.
Most large and medium enterprises are still primarily dependent on in-house infrastructure for their IT infrastructure needs. Some of them do use Amazon Web Services (AWS) for specific cases such as running a web-based production service. But AWS often doesn’t meet their security, compliance or performance needs when it comes to things like developing intellectual property, crunching large amounts of internal data or developing new applications. Also, at scale, AWS can actually be quite expensive. When enterprises need to crunch large amounts of existing data, they often end up having to do it in-house. All of this leads me to believe that private infrastructure is not going away anytime soon.
Similarly, most startups initially use public clouds (AWS) to meet their infrastructure needs because of the variable cost model and the ease of use. However, once they get to the ~100 employee mark, the costs start to add up, and they often face the myriad performance issues associated with a shared infrastructure. They then begin to explore private cloud options.
We are starting to see CIO organizations trying to build out the private cloud on their own as a way to provide better services to end users and to reduce operational costs. But down the road, more often than not, they run into trouble. Some of our portfolio companies at Foundation Capital spend over a $1M+ per month on AWS and often yearn for a complementary private solution for software development and testing. Some have had a dozen engineers focused on their OpenStack cloud, and yet, it took them more than six months to get it up and running. They considered all the existing vendors in the space and found them to be consulting companies with large armies of integrators, who benefit from keeping keep OpenStack complex and brittle.
A recent IDG survey indicated that 84% of CIOs expect to use OpenStack as part of their private cloud software stack. Large, medium and small — companies of all sizes are interested in the private cloud. While AWS continues to gain traction, the market for the private cloud is at least 10x that. But existing private cloud solutions are still not enterprise grade. There still needs to be a solution that actually fixes the computer science issues — and is a complete turnkey to manage and operate.
At Foundation Capital, we believe that the cloud is neither an integration project nor just a system, but rather a system that has to be architected and operated like product. By integrating individual components, one misses the broader view that a performing, well-designed product cannot result from basic integration.
We strongly believe that enterprise private cloud, next generation virtualization using containers, and warehouse scale operating systems are filled with opportunities for the next new billion dollar enterprise IT companies.