keywordsSoftware Defined Networking (SDN)
Software-defined networking: hype or reality?
Date: Fri, 12/14/2012 - 17:38 Source: By Manek Dubash, journalist
Software-defined networking (SDN) has been riding up the hype cycle for some time now, according to Gartner research VP and analyst Ian Keene. So what is SDN, and is the hype justified?
Manek Dubash, journalist
Image credited to NetEvents
Today's datacentre networks use switches and routers that are effectively two systems in a box.
The first system consists of the electronics that move or manipulate packets, and is known as the data plane. The second is the control plane, and consists of those rules which decide where packets should be routed and how they should be manipulated. Some of those rules maybe hard-wired into the device, some may be configured by the network administrator.
SDN separates these two functions so that the control plane is sited centrally in a server. This runs applications that control the switches over the network using a new, open standard protocol called OpenFlow. The switches then carry out those instructions, becoming little more than packet routing and manipulation devices.
The organisation tasked with guiding this technology is the Open Network Foundation (ONF), which keeps the network vendors off its main board in order to ensure that its approach is non-partisan.
The advantages of SDN, according to the ONF, include its ability to centralise network policy management and implementation, so reducing costs and complexity, and crucially, to provide flexibility because the system is designed to be open, to allow third parties to develop new applications for the network. These can provide packet routing instructions, or add security to the network that's tailored to the organisation, for example. This means that, if you need an application that isn't in the box, you can write one or get one written, without having to wait for the vendor's next product development cycle.
Keene said the advantages include: "Quick service deployment; enabling you to deploy new services because you could do policy-based control here. And you can analyse also what is going on and make changes as necessary. You can provide infrastructure as a service much easier. And you can provide the quality of service needed, on demand."
That the idea has been very quickly been accepted by almost every switch and router vendors speaks to the likelihood of it becoming reality.
As Keene said: "Companies such as Huawei or Alcatel-Lucent, if they are going in to pitch with a leading service provider, have got to have a software-defined networking story. They don't have to have a product yet, but they have got to have a story. And, in theory, OpenFlow will make a multi-vendor switch environment much more of a practical reality."
The vendor angle
Most vendors have been keen to talk about SDN, not least because it gives them a potential lever against the switch market's leader, Cisco.
Dell's VP of product marketing Arpit Joshipura said that the choice for customers was wider than just SDN or not SDN. He said there are overlay hypervisors such as those from VMware and Microsoft which manage the pipe from the outside, and will need a layer 3 switch. Second are the legacy vendors who, he said, will drag their feet because they have the most to lose, with 20 years of investment. Third are greenfield sites, especially in academia with no incumbent, and who can afford to can take more radical approach. He also said that there are a whole set of standards that have yet to evolve with SDN.
Markus Nispel, chief technology strategist at Enterasys, commented that SDN is here to stay, and is not just riding the Gartner hype cycle only to eventually disappear. He also disputed the ONF's claim that SDN emerged from Stanford University, saying that Cabletron, as Enterasys was once called, was looking into the idea in the 1990s.
Nispel added that the advantages of SDN were many: "It goes beyond automation, [...] more towards the orchestration of different IT services including the network infrastructure to quickly deploy new services. We see SDN being a very viable approach for carriers, and for very large data centre operators as well. And we do see SDN very valuable for enterprises, but in a different fashion."
Shehzad Merchant, VP of technology at Extreme Networks said that SDN had arrived to solve the problem of stress on the datacentre network caused by the deluge of data from BYOD, mobile devices, cloud computing, and the opening up of the network to a wider audience.
He added that the switches will not just be dumb devices: "SDN only emphasises the fact that the network has to offer higher performance, higher capacity, and lower latency. So there is a fair amount of innovation to be required at the networking layer itself."
In contrast, HP's Mike Banic said the real value in SDN is the applications. As VP of global marketing, Banic said: "The big topic everybody's focused on so far in the datacentre has been virtualising the network. Well what about doing other things in the network, like automating security or providing other network services like load balancing or WAN optimisation that are delivered today through dedicated appliances? And those dedicated appliances themselves present limits in terms of scale."
Speaking for IBM, business executive Charles Ferland said the key thing to remember was the SDN is all about the software, and that the real beneficiaries of SDN would be cloud providers, who would be able to automate many of their processes, and service providers.
"For them the potential is absolutely huge," Ferland said. "From now on they no longer just provide the pipe, but they have a view on every single flow going on."
For example, he said, it means you will be able to detect fraud detection, and other security threats. Ferland said that providers could also detect traffic patterns over time and pre-emptively sell additional bandwidth to customers before they needed it. SDN could save money too, although like any technology, how much will depend on the care with which network architects plan and implement their systems. It may even allow datacentre managers to do away with additional appliances such as firewalls, Ferland said.
There are challenges, according to Nispel. He said that scaling an SDN network could be a problem, especially for service providers with very large networks and lots of applications hosted by lots of customers.
Not Cisco? Good
Among network vendors who are not Cisco, there is a groundswell of support for SDN, which is seen as disruptive, and therefore opens up opportunities to grab market share. Cisco should not however be discounted: in a recent speech, the San Jose company's CEO John Chambers said that while Cisco is behind SDN, it will not happen overnight - the inference being that Cisco plans to hang onto its share of the datacentre network market.
For the rest, however, there are opportunities, so end users can expect network vendors to be holding out the promise of easier, cheaper more flexible networks, powered by SDN technology.