5 steps to a software defined network

sdn cover orig
Credit: Tavis Coburn

Disclaimer: Zeus Kerravala is an employee of ZK Research and his clients include Cisco and Avaya.

Earlier this month Avaya held its annual Executive Partner Forum in San Diego. As expected, one of the hot topics at the event was software defined networking.

For Avaya resellers, SDNs provide an excellent opportunity to gain share on Cisco and the rest of the field. In the 2015 ZK Research Network Purchase Intention Study, 40% of organizations are willing to look at alternative network vendors for their SDN solution. With traditional network equipment, only 10% are willing to look at swapping out their incumbent. In networking, SDNs are the biggest, best opportunity for share gain.

Lets be honest, Avaya is still not top of mind when customers think of networking, although Avaya has an excellent SDN solution called Fabric Connect. I’ve often called the Avaya Networking product the best kept secret in the network industry as the quality of product is far stronger than its share reflects. Prior to the event I had an opportunity to interview the architect of Fabric Connect, Jean Turgeon, (commonly referred to as JT) VP and Chief Technologist, Software-defined Architecture, WW Sales. Despite being a Montreal Canadiens supporter, I find JT to be one of the sharpest people in the networking industry.

JT and I discussed how the marketing hype around SDNs has outpaced deployments and together we brainstormed five steps that organizations should take to deploy an SDN.

1. Simplify the environment. Traditional networks are filled with legacy constraints that make the network difficult to manage. Over the past two decades, more and more network protocols have been added to improve the performance of the network. For example multi-link trunking (MLT) was added to increase resiliency while Open Shortest Past First (OSPF) was developed to increase the dynamism of routing.

Also, many of these protocols are now intertwined so if there’s a bug at layer 2 then all the upper layers may also be impact. Network professionals should look to eliminate as many of the protocols and dependencies as possible by evolving to a protocol, like shortest path bridging (SPB) that was designed for the cloud era.

Part 2 of simplification is moving away from “active-passive” architectures to “active-active”. The concept of active-passive is old and outdated and requires having to overbuild a network. With active-active, traffic can be distributed across the different links increasing both performance and uptime. JT and I both agreed that the first step was the most important. Making any kind of change to a complicated system is going to increase the level of complexity.

2. Deploy a network fabric to enable edge automation. With a traditional network, implementing changes means touching every network device. This is long, laborious process that is highly error prone because of the repetitive nature of manual configurations. ZK Research has found that, on average, it takes four months for businesses to implement new network services – far too slow for the digital era of business. Network fabrics enable provisioning to be done at the edge of the network only with updates to the core being automated.

JT told me that the automation capabilities of Fabric Connect enable Avaya’s customers to enjoy an 11x improvement in the time taken to get new services to market. I’m always skeptical of claims like this but JT directed me to an independent report from Dynamic Markets Limited that provides a number of case studies from customers that have deployed the technology, supporting the data point and providing others. A good read for any customer looking to automate network configuration.

3. Enable application automation. No one makes changes to the network just for the sake of making a change. Typically network changes are tied to applications changes so it makes sense to automate this process. For example, if a business is going to conduct a video call, the video server should have the capability of directing the network to create a dedicated connection between the two points. Then, when the call is over, the application can remove the path. This works most effectively with an end to end solution that includes application software, network infrastructure and developer tools.

As an example, Avaya can accomplish this with a combination of Aura, SDN Fx and Engagement Developer Platform (EDP). JT shared with me, they will be demonstrating how both SDN Fx and EDP come together within a single EDP workflow at the upcoming Enterprise Connect show. Looking forward to seeing the demonstration.

4. Enable orchestration.The data center is more than just a network. In addition to configuring the network, the applications should be able to control the configuration of other infrastructure such as storage and servers. Initially, this is best done through “integrated stacks” that remove much of the complexity of tweaking and tuning the specific configurations of the underlying technology. Avaya has an automated network where IP is no longer based on a distributed routing hop model. Instead it is a service running on top of the fabric, hence IP uses Ethernet Switch Paths and eliminates IP “hopping”, making this a solution that should be considered by any company looking to simplify network operations.

5. Enable Open Orchestration. SDN nirvana would be defined where applications can orchestrate the deployment and configuration of everything including firewalls, IDS/IPS, load balancers, white label switches and other infrastructure. Also, it would work with physical and network functions virtualizations (NFV). This would necessitate the need for an open SDN controller that speaks something like OpenFlow. Other gateways would be required for protocol interoperability such as VXLAN to SPB. Avaya currently does not have an SDN controller.

I asked JT why Avaya is lagging behind in this area. He said Avaya chose to take an alternative path to SDN. Instead of shooting for SDN nirvana today, the company has focused on building a strong underlying foundation that provides benefits now but can be a platform to build a scalable SDN solution on down the road.

With SDN I strongly recommend that customers not make a decision based on who the incumbent is and evaluate three to four vendors. Unlike traditional network infrastructure, where most of the products are similar to each other, SDN products can vary greatly. Avaya’s solution is a good example of a differentiated solution that should be part of that pool of three to four vendors.

Achieving the Utopian state of SDN is likely at least five years away for most businesses but it is the goal that organizations should be striving to achieve. The first step, and JT and I are in agreement here, is to simplify and automate the network. One can’t achieve any benefits with a network that is complicated where it takes months to make even the simplest changes.

To comment on this article and other Network World content, visit our Facebook page or our Twitter stream.
Must read: Hidden Cause of Slow Internet and how to fix it
Notice to our Readers
We're now using social media to take your comments and feedback. Learn more about this here.