HP: It’s Time For Users To Question Cisco’s Role

New technologies underneath

But all this needs a network that handles content better, with an awareness of its ultimate use, said Congdon, who wants to see the end of the requirement to distribute multiple versions of content. “It’s more optimal to do the transcoding at the edge.” That plays well to HP’s Open Network Ecosystem (ONE) which hosts applications on blades in the HP switch chassis.

Future network equipment will move to a new level of virtualisation, he predicts, given advances in silicon and two particular developments: chip level optical networking, which will increase the speed of network components, and memristors, which Congdon predicts will increase the capacity of solid state storage by “a couple of orders of magnitude, offering terabytes on a thumb drive,” because they store values other than one or zero.

All this allows better price-performance at the network edge, and through I/O virtualisations could make hardware more flexible. “Today there’s a whole layer of software which makes it possible to run multiple virtual machines on the hardware. What if you could reconfigure the hardware itself, and move it down to a lower layer?”

A new generation of hardware, in a few years’ time, could remove the software hypervisor layer, and use the newly flexible physical hardware to create a physical machine that matches whatever the application needs, predicts Congdon. A fully disaggregated server with optical chip interconnects is “some years off”, he says, but it could allow the same hardware to turn from a network switch to a blade server, he said: “That should result in reducing cost for customers, driving out cost and driving towards standardisation.”

Too much equity goes to too few players

But the biggest barrier to this could be leaving Cisco in a position of dominance, he implied. “Too much equity has been going to too few vendors,” he said, in the first of several not-very-veiled references to the network giant.

Buying Cisco equipment could actually delay innovations, he suggested. “Because too much equity is going to too few people, customers are having to spend more than they need to. Companies can’t spend on their own innovation, because they are spending too much on network vendors.”

Cisco’s dominance in most markets is “way out of whack,” he said. “People should be getting a better deal, they should be opening their RFPs [requests for proposals] to more vendors”. The combination of HP and 3Com could be big enough to persuade users to do that, he said.

A lot of developments in the industry are “crazy” because they have to work around what Cisco does, he said: “A lot of people are coming up with crazy architectures, becuase they have to work around the gorilla. People are doing things because they are trying to work around the  incumbent architecture.”

This includes things like cloud architectures, which overlay network intelligence in a hypervisor – “because they can’t work with the network itself”.

Instead of getting better, the situation actually gets worse, said Congdon. While its dominance continues, “the incumbent keeps coming up with new proprietary architectures to try and defend it. They make it way too complex, so everyone has to be a rocket scientist.”

Networking could actually be a lot easier to do if the “incumbent” [Cisco] wasn’t protecting its business model, he said: “The industry isn’t competitive enough and isn’t standard enough. We aim to change that, obviously!” That means “pushing for a more standard network architecture,” and in the process, HP Networking hopes to become number one in networking, not number two… at least Congdon says so.