When it comes to Silicon Valley, it is the venture capitalists that are the big-name Hollywood producers. Convince one of them to get on board, and your dreams of digital riches could come true. That was certainly the case with .NET on Linux and mobile company Xamarin, one of Max Gazor’s first investments when he came to Charles River Ventures in 2011.

Since that time, Xamarin has become wildly successful, and Charles River Ventures has found other companies to invest in. We sat down with Gazor to discuss the future of software in the enterprise, and to gauge the current investment climate for enterprise software companies.

SD Times: Big Data was a big deal for VCs. Is it still?
Max Gazor: We were there at the first wave of analytics companies with Netezza. It was a convergence approach to data warehousing that was faster and had a better price/performance ratio. Hadoop is the other opposite extreme: big and large pools of data, but maybe not as high value. It’s moving into the space of data warehousing. The investment in Hadoop really happened in the 2008 and 2009 timeframe.

We’ve seen a considerable slowdown in the infrastructure behind Big Data. Cloudera and Hortonworks are dominating as distributions. They are owning the channel. It’s harder and harder for new startups with wizbang infrastructure products to go into that market.

Most of the time, when companies come in today, they almost all have a token slide on their Big Data infrastructure. It’s no longer a differentiating factor. I don’t necessarily think the Holy Grail vision of replacing application infrastructure with Hadoop on the back end is going to look revolutionary. It’s not going to look revolutionary; it’s going to be evolutionary. You build layer upon layer of infrastructure, and once you have the back end, then you need an identity layer, a compliance layer… I think ultimately the vision of running an ERP on Hadoop infrastructure is getting there, but I don’t think it will happen soon.

Do you see any companies with vertical-targeted Hadoop-based applications?
I don’t necessarily see SQL-based databases going away. Hadoop, as an application database, is nowhere near ready for that. It’s still very much an analytics/query type thing. If we see a company today targeting a vertical, and it’s mining a lot of data, there’s going to be a segmentation of databases.

For their Web application, they need something responsive, but yet every night they run huge batch workloads. That gets them a lot of analytics. Hadoop is not a replace-all or catchall. What you’re starting to see now is segmentation around where—before we had Web apps and analytics apps—now you’re starting to see segmentation around real-time apps and memory-centric stuff.

How will containerization change the Hadoop landscape?
Hadoop and virtualization were never consistent, in the sense that when you think about it, what’s happening is that Hadoop was written in a time when network was the big latency component. You’d have these huge delays. But now, the network fabric is so fast you don’t need to bring compute close to the data. You can operate on large distributed environments, like a cloud.

Then, if you play that forward, local processing and memory is going to be a greater factor over time. The scheduling of stuff is no longer the bottleneck. Virtualization was never a good fit for Hadoop. What you’re starting to see now is this emergence of containers and running on bare metal. I predict that as containers take over, that you’ll find better info and better hardware constructs to support bare metal with containers.

You don’t need this overweight hypervisor that can run containers. The natural use case for that could be Big Data. When you’re running on a hypervisor you’re paying a tax.

Containers are a nice solution for the middle. One of the big hopes for containers is Linux applications, and running large distributed, stateful applications that are non-Windows. Big Data is a great use case for that. We’re kind of in this post-Windows, post-hypervisor world. Big Data clearly makes sense to run on containers.