10 things we learned – or didn't -- from Cisco's Insieme launch

With Cisco/Insieme debut some things are clear but others are still open questions


Cisco's introduction of the Insieme Networks product line and strategy was the most anticipated networking announcement in almost two years. It represents Cisco's response to the software-defined networking trend pervading the industry, and threatening Cisco's dominance and profits. While not a watershed moment – Insieme didn't change Cisco's modus operandii, as explained below -- Insieme is nonetheless one of the most important events and product introductions in Cisco's history. Here are 10 things we learned or have yet to learn from the Insieme launch:

• Cisco takes the SDN threat very seriously – Otherwise, they wouldn’t pay up to $863 million on an 20-month-old spin-in company to develop its next-generation data center product line in response to SDNs. By comparison, the ASR 1000 router cost $250 million and took five years to develop.  

• Cisco is still a hardware company – Why else would they respond to a software trend and threat with hardware? Despite all of its stated plans to become more of a software company and increase its software revenue, Cisco’s business model is hardware driven. SDNs de-value Cisco’s hardware.

[RELATED: First look: The Cisco/Insieme family]

[THE WAIT IS OVER: Chambers: Cisco waited too long to address SDNs]

• Cisco is intent on protecting its installed base – The standalone version of the Nexus 9000 – open source controller extensions, merchant silicon – is for DIY users. Customers leaning toward a completely integrated system are targets for the custom ASIC-based Nexus 9000 with the Cisco-developed APIC controller. Enterprises will likely adopt this platform vs. the standalone version, which might appeal to cloud providers considering white box hardware and open source software. Cisco will eat its own before allowing a competitor to.

• Cisco is intent on protecting its profit margins – Another reason Cisco funded and responded to SDNs with Insieme hardware. Cisco enjoys 62% gross profit margins on its products. Margins on merchant silicon-based white box switches are one-third to one-quarter that. If you are a Cisco customer and you want a programmable network, Cisco will strongly encourage you to buy the Application Centric Infrastructure (ACI)-mode Nexus 9000 hardware.

• Cisco believes the physical/virtual approach to programmability is the best – Cisco argues that software-based overlay approaches to SDN are optimized for provisioning virtualized applications and resources, while data centers have both virtualized and physical resources. Those physical resources include the networking infrastructure itself. Cisco claims its Application Centric Infrastructure, which is founded in the new Nexus 9000 hardware, can capably provision both physical and virtual resources in a data center.

• Cisco believes the software-based overlay approach is too costly – VMware disagrees, but Cisco says when upgrading from 10G to 40G, ACI and the Nexus 9000 infrastructure reduce TCO by 75% over a merchant silicon-based infrastructure running a software-based network virtualization overlay. Included in that calculation is a “per-VM tax” levied on users of those merchant silicon switches running a software overlay that accounts for most of the cost.  This is a key point in Cisco’s argument against the software overlay approach and in its data center networking competition with VMware.

• Cisco still has to partner with VMware for server virtualization – Even though Cisco and VMware are becoming intense rivals in networking – thanks to VMware’s $1.26 billion purchase of network virtualization start-up Nicira – the two still have to partner. VMware is the leader in server virtualization and Cisco is the leader in data center networking, which makes it incumbent upon both to buddy up with the other. That’s why VMware is a partner in the Insieme Nexus 9000/ACI launch – and why Cisco is not for VMware’s NSX network virtualization platform.

• No future for FabricPath? – Cisco won’t admit it publicly but, why would there be? The company’s investing up to $863 million in ACI as its strategic fabric, which is based on VXLAN; FabricPath is based on TRILL. ACI is its response to SDN, which wasn’t quite in vogue when FabricPath debuted over three years ago. When we asked Soni Jiandani, Insieme senior vice president of marketing, if ACI’s introduction obsoletes FabricPath, she gave a noncommittal “I don’t think so” before explaining that Cisco has many customer requirements to meet with its myriad data center switching platforms – Nexus 2000, 3000, 5000, 6000, 7000, and now 9000. The Nexus 7000, however, upon which FabricPath resides is now being marketed as a fabric “pod” and as a data center interconnect after the ACI launch. Maybe it’s us, but “pod” sounds like a smaller scale limited deployment. And FabricPath and ACI would not co-exist in the same fabric, said Insieme founder Mario Mazzola. Anyway, Cisco says it has an evolution path for FabricPath customers to ACI.

• Is there a data center future in any previous generation Nexus hardware? – The Nexus 7000 is now a pod and data center interconnect. The ACI Application Virtual Switch that debuted with the overall ACI architecture and Nexus 9000 switches is an adapted version of the Nexus 1000v Virtual Ethernet Module. The adaptations include reliance on the ACI APIC controller for control plane functionality, not the Nexus 1000V Virtual Supervisor Module; and support of the ACI policy model for integrated physical/virtual visibility, traffic steering and common management through APIC. Cisco also plans to extend ACI with Open vSwitch as part of its intention to open source the ACI policy model. The Nexus 2000 fabric extenders are supported in ACI. The Nexus 3000 seems targeted specifically at ultra low-latency, high frequency trading. The Nexus 5000 will be used for Fibre Channel and Fibre Channel-over-Ethernet applications, but these are also expected to be added to Nexus 9000 ACI leaf switches. The Nexus 6004 is a 40G, 4RU switch that Mazzola says could be “leveraged” as spines in the ACI fabric architecture. Could the 1RU 10G Nexus 6001 be leaves then? Cisco says its customers have many requirements that necessitate many different form factors to satisfy.

• Maintaining dominance in networking is vital to Cisco’s IT ambitions – Cisco’s goal in IT – to be the No. 1 IT company in the industry – is rooted in the network, where it is dominant. Any incursion into Cisco’s data center networking bedrock, be it from VMware, Arista, white box, HP, Juniper or anyone else, is not only a threat to that foundation but to Cisco’s broader IT ambitions. Connectivity is Cisco’s game. Servers and storage and virtualization all hang off of that. Any disruption to its core is also a disruption to its periphery. ACI and the Nexus 9000 is intended to thwart any potential disruption and clear any obstacle in Cisco’s path to becoming the No. 1 IT company. 

Jim Duffy has been covering technology for over 27 years, 22 at Network World. He also writes The Cisco Connection blog and can be reached on Twitter @Jim_Duffy.

Copyright © 2013 IDG Communications, Inc.

The 10 most powerful companies in enterprise networking 2022