It isn’t easy being a start-up. First, the SDN future looked bright, promising start-ups an opportunity to unseat incumbent IP and Ethernet networking titans. Concerned network and cloud service providers had banded together to form the Open Networking Foundation (ONF.org – now at 100 members) to help define programmable networks with a separate, open control plane, initially based on a coincidentally like-minded university’s OpenFlow controller for an enterprise networking test bed. Facing the doom of rapid traffic growth with little new revenue (the classic ‘scissor’ diagram), the current NSP Capex model to scale networks needed to change or they would need to raise prices, potentially strangling the life from the mobile and Internet golden goose pair. And to take server virtualization to the next level, namely VM mobility, Cloud SPs need network virtualization and SDN to make them bandwidth dynamic and enable application calls for bandwidth on-demand. So venture capital flowed like milk and honey to eager SDN start-ups, all expecting to emulate Nicira’s very good fortune.
But then the sleeping networking giants (a.k.a good shepards, or jolly ranchers?) stirred when they saw their cash cow herds at-risk to the new start-up wolves. Should they armor their cash-cows or might a scarecrow do? On April 8th, 2013 OpenDaylight was launched. From ODL’s mission: “At this early stage of SDN adoption, the industry acknowledges the benefits of establishing an open, reference framework for programmability and control through an open source SDN solution. Such a framework maintains the flexibility and choice to allow organizations to deploy SDN as they please, yet still mitigates many of the risks of adopting early stage technologies and integrating with existing infrastructure investments.”
That certainly sounds like a start-up/VC scarecrow to me, and portends some lipstick for the switch and router market cash cows.
But wait, there is more...
During on-junipere28099s-june-11th-analyst-day/index.html" target="_blank">Juniper’s June 11 Analyst day, CTO and Founder Pradeep Sindhu explained why x86 GPUs are inefficient for L1-3 processing compared to ASICs like Juniper’s Trio chipset. I did check with Intel, who tend to agree on GPUs, at least for the foreseeable future. Juniper’s Contrail acquisition was also mentioned as busily retooling Juniper products for SDN behind a velvet curtain.
Then at Cisco Live, “Internet of Everything” touting CEO John Chambers announced Cisco was spending $250M for its next generation routing ASIC that can connect everything that can be connected in the world. The underlying message is don’t worry, IP routing is infinitely scalable, so you don’t need to rush to find cheaper switching alternatives, Cisco’s got you covered, good night, sleep tight, we’ll wake you when it is time to buy. Cisco also hinted that Insieme is developing a converged god-box for switching/routing, compute and storage. Convergence in a box (or blade center or rack) with external control seems to be the Inieme direction take-away.
So the ‘good shepards’ have more than just OpenDaylight to defend their positions. Although others disagree, I am not yet a believer that ODL is a just a front to delay SDN progress, but my ODL update briefings do keep getting rescheduled - curious. I really do want to believe ODL will deliver an open supporting framework and code for open SDN and I don’t see that as a threat to incumbents' businesses.
Side-bar: Apart from SDN and NFV architectural-oriented efforts, I see two distinct ICT infrastructure equipment options emerging from the SDN tempest:
1-Virtualized compute + network + storage on general purpose distributed data center hardware (a.k.a. Cloud). Also able to run virtual network appliances and infrastructure, even on an hourly charging basis. For a preview, see Amazon’s Web Services Marketplace. There you will find a variety of enterprise class appliances for WAN optimization, ADC, Network monitoring, APM/NMP, etc. all hosted in AWS’ multi-tenant cloud. VMware and others also have similar marketplaces and offers. Does this sound like it could be the new revenue sources you’ve been looking for, Ms. Telco?
2- A converged ‘god-box’ for compute, switching/routing and storage, able to run network hypervisors and appliance instances. Perhaps something similar to Simplivity OmniCubes or VCE integrated racks? See ESG’s Integrated Computing Platforms market landscape report for ICP details. The drivers for service providers appear to be not putting all their critical eggs in one general purpose, oversized basket and perhaps rapid reconfigurability. SDN architecture and open source components such as controllers, management and basic network virtualization functions can take care of federation and orchestration.
Back on the cash cow ranch, how can the now starved wolf-pack of SDN start-ups get past the big guys’ SDN defenses or augment the big guy strategies for a happy-ending acquisition? Since start-ups are by definition motivated and nimble, flip to the other side of the SDN coin – NFV, and its currently greener pastures. Then flank the big guys in ODL with a stealthy, exclusive NFV vendor association committed to prototyping service providers’ NFV wishes - brilliant.
So Cloud NFV was launched this week under the leadership of former programmer, consultant and ICT trade pundit Tom Nolle and his CIMI Corp. along with six yet unnamed participating start-ups. Tom notes that large incumbent vendors were not invited to the party. CloudNFV’s goal is to prototype NFV systems being defined by the European Telecommunications Standards Institute (ETSI) NFV Industry Specifications Group (ISG) effort that numerous big network operators are driving. See www.etsi.org for the January 22, 2013 NFV ISG launch press release, NFV vision white paper and list of NFV ISG members. Tens of billions of annual service provider networking gear spending can go back in play, split between legacy equipment, virtual big iron (1 above), converged ‘god boxes’ (2 above), plus software and services.
The moral of this SDN/NFV story is the customer is always right, and their suppliers just can’t nod yes forever if they hope to keep them as customers. Customers’ business needs to be earned every day. Old-timers might recall when network operators promoted voice soft switches, external control plane (SS7) and the IP operating system (IMS), and with their help, were able to build mobile networks. However network operators were mostly caught by surprise at the amplitude of the Internet’s boom, but were ‘rescued’ by the likes of Cisco and Juniper, but now it is time for the operators to take back control of their destiny. The start-ups and other non-IP/Ethernet systems vendors are willing and able to help them deliver more bits at lower cost. The most likely next move by the incumbents is to expand ODL to include NFV prototyping, or could it be something else?