There's no shortage of new company ideas being tossed at venture capitalists these days, but that doesn't mean many of them are going to get funded. Burned by lots of network flops over the past year or two, venture firms have become more careful about which companies they fund. Foundation Capital's Adam Grosser, a general partner, recently spoke with Network World News Editor Bob Brown about which technologies have him most intrigued.
Q:What's the crop of prospective network startups looking like?It's hard to find companies that are unique these days. I've looked at over 100 projects since the beginning of the year and have not funded anything, though that's pretty typical for us -- we only do two projects per year per partner. I have not seen anything even remotely worth funding in the carrier space and given the state of the carriers, I'm focusing a lot more on the enterprise.
Q: So what's hot in the enterprise network market?The next big push is going to be in data center automation. There's a lot of opportunity to do work in InfiniBand switching, intelligent clustering and smart failover. There's a lot of stuff that will capitalize on new bus technologies and server architectures, particularly the blade servers. So I think what you're going to see is the replacement of the Ethernet switch at the top of the stack for these data centers that have north of a thousand individual CPUs.
You'll see what I'd call "intelligent controllers", things that have some concept of management. What you find when you look at some of these companies, like some of the genomics companies, is that they'll have two or three thousand 1U servers racked and stacked and the way they manage them is guys walk around and look at the green lights blink. So it's not at all well conceived. There's a software play here and an embedded hardware play. And when you talk to the CIOs, there's major pain around this, so I think that's going to be a great enterprise opportunity in the next 24 months.
Q: Speaking of InfiniBand, what's your take on it?It's a pretty good internal bus architecture but I'm not convinced yet that it plays outside the box as the interserver interconnect some people would like it to. There's an opportunity and a challenge. The [Remote Direct Memory Access] capability is very cool because it allows you to virtualize your CPUs. The challenge is that rewriting your applications to take advantage of that is a lot of work. Really the only major application vendor I've seen do that is Oracle. The jury is still way out on this.
Q: So what other enterprise network areas are attractive?There's an interesting play around wireless data management and all the issues around security and access and roaming. Another broad category is IP VPN replacement. I've looked at four such companies in the last month. People are using SSL or HTTPS and some interesting software to provide remote access for employees, for partners and for a broader sort of a consortia approach to a project. This is without using a VPN, where the access is a little bit more ubiquitous and it's not coming in at the Layer 2 level of your network.
Q: What about the category of route optimization companies, such as RouteScience [Grosser sits on the board], NetVmg and Sockeye?Yes, it's a legitimate category. Route optimization is to routers what load balancers have been to servers. BGP is broken and I don't think it's going to get fixed anytime soon. Now, whether Cisco should own [this area], that's another story I don't know the answer to. The IETF ought to fix [this problem ] and has tried for the past eight or nine years, but there's so much politics and B.S. in the IETF that I don't think it's going to happen anytime soon.