What Should a Revamped Cisco Look Like?

"The big ones they've made lately... none of those as a matter of fact drove router and switch sales," Kerravala says. "The last one that in fact drove switch sales was Selsius," a VoIP (Voice over Internet Protocol) company Cisco bought in 1998.

And again, they should enter fewer market adjacencies and only those that drive the enterprise and service provider routing and switching business, Kerravala says.

"What Cisco needs to decide is what are the adjacent markets they want to go into," Kerravala says. "I would think there'd be less emphasis on things like direct consumer markets. They need to get back to paying attention to the stuff that made them as a matter of fact successful, and that's the channel-driven business sales force."

Switching revenue was down 7% in Cisco's second quarter and profit margins missed the mark as a new generation of Nexus and Catalyst switches with better price/performance cannibalized sales of others. Analysts say Cisco needs to better align Nexus and Catalyst operations to avoid a similar fate down the road.

"Part of the problem with Ethernet switching is the multiple groups involved: Catalyst, Nexus, Nuova, MDS, etc.," add Oltsik. "It leads to product issues like the Nexus 5000 not supporting FabricPath."

Cisco as well recently introduced a Nexus switch - the 3000 - based on merchant silicon instead of internally-developed custom ASICs. The Nexus 3000 is based on Broadcom's Trident chipset. In the meantime, Broadcom is adding Cisco's AdapterFEX data center fabric switching innovation to its silicon in what some observers believe is a quid pro quo arrangement between the two - Cisco uses Broadcom silicon if Broadcom adds Cisco research as an alternative to Cisco developing its own custom ASICs.