It's Gold Rush time again thanks to VMWare's proposed $1.26 billion acquisition of Nicira, and I have seven observations on the deal.
SAN JOSE – It's Gold Rush time in California again thanks to VMWare’s proposed $1.26 billion acquisition of Nicira Inc., a startup pioneering software-defined networking. There’s no doubt there’s gold in them thar hills, but it’s too early to tell how much and who will drag the bullion to the bank.
I have seven observations after spending a good chunk of last week hearing commentary on the deal. In short, there is gold here, but how much is a matter of some debate. There’s pressure to jump in the stream with engineers among the first getting wet. Expect some industry realignment before we are done. Cloud computing plays a substantial role. And there are more loose ends here than in a stack of denim at a Levi’s factory.
First some background.
Software defined networks (SDNs) take network jobs run in protocols supported by router and switch ASICs and turn them into server apps. The goal is to shorten the time and effort it takes to set up and manage networks.
Stanford University got the ball rolling on SDN with its Clean Slate program in 2002. By 2007, three of the leading thinkers in the field were creating a vision of software-defined networks, an anti-protocol interface called OpenFlow to implement it and a company called Nicira to create the software products based on it.
The people and their ideas were compelling enough to attract $50 million in venture capital. Late last year Nicira announced its Network Virtualization Platform along with five customers for it–AT&T, eBay, NTT, Rackspace and Fidelity Investments.
Meanwhile VMWare, grown big and hungry as a supplier of virtualization software for servers, has been trying to extend its reach into the cloud. Bigger rivals such as Hewlett-Packard and IBM were threatening to eat its lunch. So VMWare snatched up Nicira and the Gold Rush began. 1. This is real
A Verizon technical manager said he has seen the glinting metal of SDN and it passed his test.
“If we continue to build networks the same way we have and project our growth into the future, it becomes an untenable business model,” said Prodip Sen, who heads a dozen-person advanced architecture group for the carrier.
“With SDN, I can separate out parts of networks, introduce new services faster and assess costs to each piece that provides a given service,” said Sen at the NetEvents Americas conference in Miami last week. “If I cannot do that, I can’t plan my business properly,” he added.
So Verizon is working with Hewlett-Packard and Intel to create a set of SDN test labs. The trio created some proof-of-concept demos last year. Now they are working on new ones based on real user scenarios with plans for field trials next year.
In this way, SDN is part of a broad effort to expand from pushing bits to owning business processes. “We are moving away from managing devices to managing services, and operators are trying to figure this out,” said Greg Gum, chief marketing and business development officer for equipment provider Telco Systems.
In a blog, Nicira investor Ben Horowitz pitches SDN as part of a 30-year change of era in communications, making the process of setting up and running networks easier and more open.
While these are "simple words", they do not tell me what SDN is doing. Are we using routers and switches? From what I have read, I still do not understand if SDN is bypassing these (which would be impossible, since they are the network paths) or what it is doing that enables it to bypass the proprietary network software. How can it bypass what is set up to only enable one way of communications? Don't these routers and switches then have to be reprogrammed to accept this non-proprietary SDN programming--are they retooled and reloaded with this, or is the original OS on there and software is then loaded on top (like old windows on DOS) to set up new means of transmission? It was also noted that SDN can tell the network there is more "network" than the routers are allowing--"applications think they have the network to themselves, when they are sharing it"--what happens to collisions and bandwidth?
The response to this would then provide a simplified answer to "What is SDN"?
The answers floating around sound like "tech bubble slight-of-hand" and every answer on the Internet looks like it was cut and pasted, with no one actually knowing what is happening with SDN, other than those who created it.
Please let me know: hippleda_at_gmail.com
No, software defined radio is quite different, as I understand it. That's more about being frequency and protocol agile by doing more radio functions in digital rather than analog blocks.
SDN is a whole new way of building networking systems. Today we make routers, switches and other gear each with proprietary hardware/software but linked using standard protocols. The SDN concept is to use more open software and API to run network jobs as apps anyone can write that run on standard PC servers.
SDN could up-end the whole communications sector and leaders such as AlcaLu, Cisco, Ericsson, Juniper etc. and their ASIC-heavy products if it takes off.
SDN moves networking functions off from proprietary hardware and software environments on dedicating gear such as routers, switches and network appliances and turns them into software applications running on open software environment on x86 servers.
SDN aims to simplify how end end users set up and manage networks (fewer interconnected boxes and protocols) and open up the task of developing cool new network capabilities to anyone who can write C programs--speeding the pace of change and flexibility of networking.