For decades now, Cisco has been the single biggest factor in driving network change. Over the years, the company has been aggressive with VoIP, PoE, MPLS, wireless LAN and most recently converged infrastructure, and has gained a significant early-move advantage. However, when it comes to software-defined networks, I think it’s fair to say that Cisco has lagged in both technology and vision, and has let the likes of VMware, Arista and Big Switch get out in front and carry much of the messaging.
Yesterday though, CEO John Chambers effectively grabbed the throats of many of the smaller, SDN pure-plays, and stated “Where are your rebel friends now?” at the company’s Application Centric Infrastructure event in New York. Specifically, the company, to no surprise, announced at the event that it was acquiring the remained of spin-in Insieme following a similar path to what we saw with Nuova and Andiamo, and went through its Application Centric Infrastructure (ACI) vision.
Getting into the details, ACI is certainly a bold vision for the industry. It promises a unified, single point of control and visibility for the management and provisioning of virtual and physical infrastructure. This would mean networking, compute, storage, virtual machines, application services and security all manage a single entity.
The benefit is that organizations can rapidly provision and migrate resources based on application policy. The thesis of ACI is that applications are the lifeblood of all companies, and having greater application agility creates competitive advantage. This means rapid provisioning, lowering the cost of operations, having the ability to make "on the fly" changes. In a sense, it brings "just in time" provisioning of IT resources to the world of applications.
In my mind, this vision is where the industry needs to focus. Much of the SDN marketing has been about commoditizing network hardware, but is that really what customers want? I recently moderated a roundtable with a number of large enterprises, and I discussed the need to reduce costs. Well, my research has shown, and this was validated by the roundtable, that all hardware costs only account for about 20% of data center TCO. Networking accounts for about 17% of that, so a whopping 3.4% of overall TCO. If SDNs indeed can commoditize the network and let’s say a company can reduce its spend by 50%, that’s a grand total of 1.7% savings. So that’s roughly a 2% savings, but at what cost? Almost everyone I’ve talked to has told me the white-box model drives up the people costs and provisioning times, which today accounts for 40% of overall TCO.
The people costs are one of the major differences between Cisco ACI and the rest of the SDN players. Cisco is focused on trying to bring the people costs involved with running a data center from 40% and cut it by as much as 75%. Customers I’ve interviewed have told me they would be willing to pay up for infrastructure if it could indeed make the overall management of the data center infrastructure easier and perhaps even automated.
I'm not going to get into the product specifics of the products involved in the ACI launch. For that, please read the esteemed Jim Duffy’s blog. He does a good job of covering the APIC (controller), new Nexus 9K switches and the optimized operating system.
I will, however, take a look at the data center partners that showed their support at the ACI event. These include BMC, CA, Citrix, Embrane, EMC, Emulex, F5, IBM, Microsoft, NetApp, Panduit, Puppet, RedHat, SAP, SourceFire, Splunk, Symantec and VCE. Why such broad support? Well, because if the ACI vision becomes reality, it’s good for the entire industry and will create a “rising tide” where all will benefit. One of the partners I talked to after the event said “this is going to be huge for us,” which was one indicator that Cisco will get strong, ongoing third-party support. It’s interesting to note though that VMware was certainly conspicuous by its absence and is clearly on Cisco’s naughty list.
However, the coup de grace of the solution isn’t the hardware, as everyone likes to fixate on, or the ASICs - it’s the “common application network profiles.” The profiles are the policies that are used to define, manage and orchestrate things like QoS, SLAs and security. The ACI-based profiles augment the existing service profiles that are used in UCS deployments. For those who aren’t familiar with service profiles, they are really the engine that drives UCS today. Talk to any IT leader that’s leveraged UCS and they’ll tell you how easy UCS is to work with because of the service profiles. Based on this, I would guess that the “low-hanging fruit” for ACI are customers that are already leveraging UCS as ACI is just an expansion of what is already in place.
Some open questions remained for me, though. While Soni Jiandani from Insieme indicated that this was not a forklift replacement and that customers could transition from existing UCS and Nexus 7K/2K environments, it wasn’t clear exactly how this happened and how simple it was to do so. I look at this as a minor point that Cisco and its partners will help customers work through, as there’s a lot to be gained by Cisco and its band of partners as the ACI vision moves to reality.