For years, Cisco has tried to sell us on the idea that its products do more than just provide a really fast pipe for the applications and data residing on the network. It calls those value-added network-based services the "Intelligent Network." In a nutshell, intelligent networks possess the ability to make decisions and provide enhanced services based on the data they're transporting.
There has been a lot of discussion from a number of bloggers, including this one from a Cisco competitor, regarding the history of Cisco's Intelligent Network strategy as it relates to virtualized networking. As for me, I wonder whether Cisco's application-centric infrastructure (ACI), the vendor's software-defined networking (SDN) play, is just more of the same, and whether it can ever move away from the stigma of a company that sells boxes that just provide really fast data pipes.
Valuable lesson on how enterprises upgrade systems
When I spent a stint as an enterprise architect with my thumb on the pulse of the infrastructure of a major government agency, I learned a valuable lesson about the buying habits of some really large network operations groups. Basically, the incumbent network vendor comes in and shares the roadmap of the various devices based on its layered approach. The network team then asks which box replaces the box that's going end of life, puts that box into the budget and buys the next-gen box.
Rarely ever is there a product that entices an enterprise to change its service model.
There's a lot of appeal to this approach when you have an existing stable network based on a vendor's existing product line. You replace the current Vendor Model 5000 with the next Vendor Model 5500 and don't disrupt your network design or network stability. Even when Cisco rolled out its Nexus line of switches, the company maintained a similar correlation between the Nexus and Catalyst lines: If you had a Catalyst 6500 series switch, more than likely you fell somewhere between a Nexus 5000 and 7000. This approach, however, creates a challenge when you're looking to deploy SDN -- even in the form of Cisco's ACI.
One of the perennial challenges with marketing the intelligent networks concept is that the network team isn't the only group that has to be sold on the concept. And they are not always the best champions for a new service model. Intelligent networks and SDN are not just technologies but new service models -- models that require changes to multiple organizations within the enterprise. Network vendors can't just walk into the office of the network manager and say, "Hey, have your developers start writing their applications using these network [application programming interfaces] APIs." The relationships just don't work this way. It's the same challenge that the network team encounters when it tries to control application behavior at the network level, as opposed to the application itself. Rarely ever is there a product that entices an enterprise to change its service model. Rather, that kind of change has to be driven by a business need.
Software-first vendors face the same challenge
This is actually the same challenge software-only vendors have coming in through the other door. VMware, for example, has a great relationship with server operation teams. Yet trying to sell NSX to the enterprise just via server teams will be a difficult challenge. Instead, software-first providers will need some attention from the network manager as well.
These marketing concerns notwithstanding, I don't see a real technical problem with Cisco's ACI strategy. Cisco makes great boxes that do great things. And no one ever got fired for buying Cisco. Cisco's challenge, in my view, is trying to convince its non-core enterprise customer that it is the vendor to implement the system and service model that ACI brings with the technology. ACI is Intelligent Network version 2.0 in every sense of the word.