Adrian Brookes, SDN technologist and director at Avaya Europe, says application centric networking is “fundamentally flawed”.
CBR Q: 20% of enterprises are using software defined networking (SDN) now, and 55% are evaluating whether to do so. What advice would you give to those 55% looking to migrate?
They should take into account the career opportunities the software defined model will offer staff. By automating some day to day tasks, they will have more time to focus on strategic initiatives and contribute directly to the bottom line. The role of today’s CIO is to make IT a strategic business lever rather than a cost centre, and SDN can help in this strategic imperative. SDN can change how services are delivered to the business, embrace agility, speed and coordination so the IT department can immediately respond to business requirements.
They should definitely consider fabric networking, as it is the best foundation on which to deliver SDN objectives. It creates a dynamic environment where new or changed services can be implemented on the fly, taking only minutes instead of days, weeks, or months to deploy. This also improves network uptime through sub-second recoveries for almost all services. Overall, fabric networking helps businesses to reduce costs and become more agile.
There are also CIOs who want to move towards greater SDN enablement but are unsure of where to start. I think the best jumping-off point is a discussion with their network and data centre teams. Together they can start building the business case for the technology and begin mapping a path to achieving their business goals, now. With the network an essential means to achieving your business strategy, an evolutionary approach is almost certainly the best way forward.
CBR Q: The SDN market is expected to reach $2 billion by 2016, up from $200 million today. To what extent is it all hype? Why?
I have seen a range of market predictions from different market analysts. While not every organisation that says it is interested in SDN will deploy it within the timescales they initially think, there has certainly been a lot of interest in SDN. At Avaya we are seeing huge increases quarter on quarter for our Fabric Connect solution.
CBR Q: SDN promises to reduce costs and increase productivity and flexibility. Are there any challenges in meeting this end and if so how can they be overcome?
The first is not for end user organisations, but for vendors, as SDN is a highly complex technology to develop. Consequently not all vendors currently have an SDN-enabled solution set available, yet they state that it is compliant to the notion of SDN. The end user customer must ensure that the vendor is following an open standards approach as this is one of the key tenants of SDN.
The second main challenge is for end-user organisations that opt for a proprietary solution. Ultimately this choice could reduce the productivity and flexibility that they gain. A solution where SDN is fully integrated into the physical network fabric provided by a single, proprietary-standards vendor, is unlikely to offer the same long-term benefits as an interoperable one. Conversely, using open standards-based products provides an easy-to-deploy, flexible solution that will work with any physical network infrastructure.
CBR Q: What are the challenges around scalability, security and interoperability with legacy networks?
Legacy networks are often cumbersome, inflexible and unstable. I hear myriad stories about corporate networks struggling to cope with the volume of traffic from today’s hyper-connected world and the dynamic pressures this places on the network and the network support staff.
Research conducted by Avaya earlier this year shows that organisations can spend more than nine months per year waiting for IT to be able to make the necessary network changes to deliver a new or improved service to the business. What’s more, over 80 percent of companies still suffer from outages caused by their own IT personnel misconfiguring changes, with each network downtime incident costing an average of £89,300. Next generation networking is a key way in which organisations can manage and overcome this issue and that is clearly of massive value to every business leader.
CBR Q: How do proprietary and non-proprietary standards fit together and how will businesses be able to work with these different visions of SDN?
Open source has been a welcome game-changer for many organisations, allowing them to cost-effectively gain functionality that suits their needs. For example, OpenStack, developed by Rackspace and NASA, has simplified cloud-based networking for thousands of organisations worldwide.
Proprietary solutions generally aren’t compatible, and as I pointed out earlier, will impede businesses who wish to work with different visions of SDN. In fact, SDN should mean that companies are not tied down to technologies created by specific suppliers, but instead can use networking technology that works with a range of different solutions, i.e. open source.
CBR Q: What’s your view on the open vs proprietary SDN battle?
We believe open standards are key to the success of SDN and we regularly contribute the various bodies that oversee the development of SDN standards and are strong supporters of this community. Reflecting this, Avaya solutions are open standards-based. Avaya Fabric Connect, for example, is 100 percent backwards compatible with existing ethernet networks and is an IEEE standard compliant solution.
Avaya is also integrating it with OpenStack’s ‘Quantum’ project. This would allow network administrators complete flexibility to move their resources around the network in a truly virtualised manner. Choosing proprietary solutions goes against the grain of many of the benefits of SDN including flexibility, agility and simplicity.
Application centric networking simply addresses the current position, where the network defines how the applications perform. This approach is fundamentally flawed as businesses are forced down one of two paths: they either provision the network as they go along, or pre-provision it en masse to cope with demand. Surely a better way to view this is: "We need to roll out this business critical application, let the network automatically ensure it provides the service suitable for the applications needs". This, to me, is a truly application led approach.