There has been no technology in networking that has had as more hype or confusion around it than Software Defined Networks (SDNs). In November alone we have seen a number of networking vendors announce SDN related products, including Cisco's (CSCO) $840 million acquisition of Insieme, and related Application Centric Infrastructure (ACI) product launch.
SDNs have been positioned as a "Cisco killer" as the technology threatens to disrupt the networking status quo by enabling organizations to build networks using commodity "white box" or low cost switches much the same way that virtualization commoditized the server industry. This is the reason we have seen so much start up activity in the networking industry lately - everyone is hoping to grab a slice of the dominant market share that Cisco enjoys today.
However, I do not believe SDNs are a threat to Cisco. In fact, I believe SDNs will actually create another wave of growth for Cisco that enables the company to expand into other markets and solidify its position in existing ones. Cisco has made a living historically from taking advantage of market transitions and the transition to SDNs is about as big of one as we have seen in networking since the voice over IP days.
Much of the value proposition around SDNs has been to lower the cost of networking hardware in the data center but I believe that is the wrong focus. I've done numerous interviews and roundtable discussions on the topic of SDNs and what CIOs are looking for is infrastructure that is easier to deploy, provision and manage rather than saving a few bucks on hardware.
The reason that IT leaders are focusing more on the operational issues is because that's the bulk of the cost of running a data center. Our firm ZK Research has shown within a data center hardware accounts for about 20% of overall total cost of ownership. Network infrastructure is 17% of that number, so a whopping 3.4% of data center TCO. So now let's use a hypothetical scenario and say that by shifting to commodity gear the organization saved 75% on the cost of hardware. This would have the impact of saving a little over 2.5% on overall data center costs.
Shifting the focus to people related costs, which accounts for about 40% of data center TCO, highlights how much bigger the impact can be. Much of this cost comes from the amount of time taken to provision infrastructure. A modest savings here of only 10% means a 4% reduction in overall costs. Ideally the organization would strive for somewhere in the neighborhood of a 25% reduction, or 10% of overall TCO, a 4X improvement over focusing on hardware alone. There is one more point worth noting and that is when commodity hardware is used, operational costs typically go up as the IT department often takes on the burden of trying to tie the hardware together through a number of manual processes. So the hardware cost savings would actually be offset by increased people costs. In many cases the increase in operational costs could outweigh any hardware savings.
I'm not saying there's no market for commodity hardware. It's been well documented that the likes of Facebook (FB), Google (GOOG), Amazon (AMZN) and Baidu use a lot of white box hardware. These companies create custom operating systems and features and use the additional functionality to create a competitive advantage. These companies also have hundreds of PhDs and Cisco certified engineers running the network. This model will have high appeal to the top 500 or so organizations with big network operations teams. The rest of the industry will look for something that's easier to deploy and utilize.
This is a big problem that the Cisco ACI solution is trying to solve. ACI is a pre-integrated, validated and tested solution that customers can deploy to automate the provisioning of data center infrastructure such as servers, networks, storage, application delivery controllers and security. Cisco's Unified Computing System (UCS) can be thought of as a mini ACI as the solution automates the provisioning of networks and servers. ACI just takes it a step further by including the rest of the IT stack. UCS has been a tremendous success for Cisco and its install base of 25,000-plus customers will be the "low hanging fruit" for ACI.
Cisco solutions typically use proprietary technology to make hard things, such as data center provisioning, easy. The solutions from Cisco typically work well enough or have enough advanced features that customers are willing to pay a premium to get stuff that works today rather than waiting for a standards based solution tomorrow. Some competitive vendors may cry foul due to the proprietary protocols but, in general, when the standards are ratified, Cisco supports them as well.
Think I'm wrong? Let's look at history. There have been many markets where Cisco has come to market early with proprietary solutions that have allowed Cisco to capture a chunk of the market before anyone else has a solution. Power over Ethernet (PoE), voice over IP (VOIP), branch routers, and wireless LANs are all examples of where Cisco has come to market with a premium price, feature rich solution early that allowed the company to grab early share and establish dominance.
I believe this to be the case with SDNs. The operational complexity in data centers is spiraling out of control and the associated people costs continue to go up as well. Through ACI, Cisco will deliver that validated, pre-integrated "SDN" solution that enables its customers to immediately cut operational costs significantly, I think by as much as 50% depending on the environment. As they have done historically, Cisco will charge a premium for the solution that customers will pay for because they get a solution that works now. Worried about Cisco margins or a drop in share because of SDNs? Don't be, it's just the latest marketing transition that Cisco will use to accelerate the business.