It's been a decade since networking has seen big venture dollars for technical innovation, but software defined networking has changed all of that.
Venture capitalists are willing to take a chance on software defined networking (SDN) because customers have demanded a new network strategy to handle server virtualization and complex cloud environments, says Mike Volpi, a partner at Index Ventures. Index was one of the first firms to put a check into the bank account of Big Switch Networks Inc., now considered an SDN pioneer and an acquisition target since its key competitor, Nicira, was snapped up by VMware Inc. last summer.
While the VMware-Nicira acquisition was a telling sign that SDN should be taken seriously, there are still risks involved in an SDN investment for both venture capitalists and end users. For one thing, the technology would take off much more quickly if network hardware vendors like Cisco would release supportive hardware, but that's not happening so fast.
Volpi, a former Cisco vice president, who once hired Big Switch co-founder Kyle Forster to work at the networking giant, explains why Index was willing to make the SDN investment.
Why are venture capitalists investing in SDN right now?
Mike Volpi: [The reason we decided to invest in] the SDN opportunity almost two years ago fundamentally came from the belief that networking as it stands today isn't working quite right for customers. SDN offers a range of unique ways in which to address those customer issues. Probably the most important [factor] was that SDN offers far better programmability of the network, particularly in the data center world. But it also has potential in almost every part of the networking world.
This is an industry that has been inside that piece of hardware vertically integrated, with every vendor working up and down the stack. The SDN application really opens up the universe.
partner, Index Ventures
In SDN, the controller essentially acts as a platform layer, which allows networking applications to operate across a multitude of different switches and networking environments uniformly. SDN also has the opportunity to more closely bring together the traditional hard switch world and the virtual switch world, which is really a software task.
Cisco has such a large hardware installed-base. Do you see SDN replacing existing hardware, or will it play into the existing infrastructure?
Volpi: It definitely won't replace the hardware. We still need those switches. SDN acts as a layer above those switches. [But currently,] applications reside as extensive software inside those switches. SDN allows network operators to abstract out [the software and applications] into the SDN layer. It allows for simplification of this very broken-up network architecture.
But isn't it the case that for an SDN layer to work, traditional vendors like Cisco will have to support OpenFlow or some other common, open standard? Is there concern in the investment community that their lack of willingness to standardize will inhibit innovation?
Volpi: Not really. The lowest-hanging fruit is the virtual switch market, which is not much of a play for Cisco. Also, there are a lot of other network vendors that do support OpenFlow. HP [Hewlett-Packard Co.] has been positive about it; Juniper [Networks Inc.] has shown keen interest; and others, including Extreme [Networks Inc.] and Dell [Inc.], have also shown interest. In the beginning, Cisco said it wasn't going to happen, but now they are saying, 'We already do that and have been doing it for a while.' I think Cisco will probably be a relatively reluctant participant, but at some point in time, they too will [support] this.
Do you believe that OpenFlow must be the protocol behind SDN?
Volpi: I don't think there is any mandate as to what specifically the protocol is. The nice thing about OpenFlow is that it is open source, so it's a level and fair playing field for everyone. The protocol itself has an interesting specification to it, but it's not rocket science. The idea of SDN is bigger than OpenFlow. Ultimately, if there are variants or newer versions of OpenFlow, that's fine. The key idea is to abstract the software out of the switch or router and put it into the server environment.
What does the VMware-Nicira acquisition mean to the rest of the SDN market? What does it mean for Big Switch, which has similar technology?
Volpi: The highlight is that the value that was placed on Nicira accentuates how real and how credible SDN has become. I think the acquisition definitely fortifies the market movement toward SDN. Obviously now VMware is a step ahead because of its commitment to SDN. But other vendors will be equally keen on supporting it, so I think it's a step forward. Big Switch and Nicira had a similar heritage and some elements overlapped, but I think they're different enough that we are confident at Big Switch about our approach.
How does Big Switch differentiate from Nicira or a newer startup like Plexxi or Midokura?
Volpi: From Day 1, Big Switch was very focused on being able to serve both the hard switch and the virtual switch market. We think that for a lot of customers, being able to bridge the two worlds is very important, and that's a critical point in the architecture and the controller. We also use true open source, which has been very successful. The product is Java[-based], which has given us flexibility, and which Nicira has done differently.
Know your software defined network startups
With Plexxi's SDN strategy, why bother with network fabric?
Midokura network virtualization: Layers 2-7 services, OpenStack
Big Switch Networks: SDN, network virtualization and an army of partners
Nicira network virtualization technology: The VMware of networking?
ADARA Networksdara launches SDN channel partner program
We have also taken a partnership-oriented strategy where we integrate into their ecosystems. Relative to the other players, our heritage out of the original OpenFlow work at Stanford, and the fact that we've spent two years developing product and working with customers, gives us a unique position in the marketplace. We have the expertise and the know-how both technically and with customers.
Looking at the SDN market as a whole, will a lot of the action be driven by applications, such as firewall management, as opposed to an overall network virtualization architecture?
Volpi: In the big picture, SDN is going to be part of multiple different parts of the network in multiple different ways. While Big Switch is focused mostly on the enterprise, a lot of service providers are interested in doing tunnelling and traffic-engineering techniques. Then you have a wide range of other applications, from load balancing to firewalling, that are thought of as enterprise or data center applications. Applications are what make the SDN market transformational, not just technically but for the industry.
This is an industry that has been inside that piece of hardware vertically integrated, with every vendor working up and down the stack. The SDN application really opens up the universe. You don't need to go to the box to do a firewall or traffic engineering. Secondarily, you can think of your network as a singular system rather than having to implement services in a box on the network. Your firewall no longer has to be placed in this little location. Firewalling becomes an abstracted process that you can manage from a policy perspective and then implement it as you see fit.
That opening of the ecosystem to app developers and the universality with which you can [direct] the services in the network make it extraordinarily powerful. That's the true vision of SDN.
Is there concern that there is a lot of activity around point products and less so around the overall architecture change?
Volpi: It's not atypical in an early-days market for vendors to grab onto different small pieces. I think as the market evolves, there will be broader solutions. At Big Switch, we focused on a universal platform and then having applications reside on it. But SDN is such a comprehensive technology that solves many different problems, you're going to find lots of companies jumping into one corner of it. That's going to happen for a while.
Will there be a shortage of SDN-savvy engineers to support market growth?
Volpi: I don't think so. This is not necessarily a situation where every engineer that used to work at Cisco can transition over, but I do think there are plenty of engineers that come from the virtualization universe that are very knowledgeable and can be leveraged to do this. Also, there are plenty of people with good solid software knowledge. Certainly from Big Switch's perspective, there hasn't been a shortage of top people that want to come work for us.