Q&A: Mike Lyons, Executive IT Architect, discusses the rise of SDN

How software-defined networking is changing the rules of provisioning and cloud governance

By | 4 minute read | September 17, 2019

Software-defined networking (SDN) plays a significant role in IT management and migrating critical workloads to the cloud. A new report underscores SDN momentum and estimates the market size will reach $130 billion by 2022. Executive IT Architect for IBM Services Mike Lyons sheds light on how SDN transforms networking and cloud governance.

What’s top of mind for companies when it comes to governance and cloud?

Lyons: For mission-critical workloads, availability and performance are critical. This puts a heavy focus on network design and capacity. Security is a concern, particularly when a public cloud is part of the solution. Clients want to protect data from exposure to other cloud users or potential attacks.

There’s also the question of what’s suitable to migrate to the cloud. One reason to store data on premises might be data sovereignty, especially as related to the General Data Protection Regulation (GDPR); sensitive personal information such as census data isn’t suitable for the public cloud.

Why adopt SDN?

Lyons: We rely heavily on skilled engineers to operate networks, but even experienced engineers can make errors. SDN can reduce human error — which is the most common cause of IT service outages — through automation.

Automation gives you faster execution speed, but every script needs to be carefully reviewed before it goes into production. With SDN, you can define automation scripts around a set of business rules; for instance, to provision a fleet of virtual machines with specific workloads and network configurations.

In a cloud environment, you stand up new functions quickly and keep them only as long as needed — this is a big change. Previously, companies invested in data center servers with multiple functions, but it’s hard to retire a physical machine if you still need a function. In virtualized environments, you spin up a virtual machine that does one thing with orchestration tools like VMware vRealize. This approach also extends to containerization.

Tell us more about SDN and business rules.

Lyons: Take firewalls as an example. Many clients with legacy environments have firewalls with thousands of rules in them; however, validating these rules is time-consuming, risky and often expensive. SDN and micro-segmentation allow you to tie rules to workloads to define network flow, so you’re not relying on a physical firewall. All the components such as virtual machines and middleware are provisioned at the same time the workload is provisioned. More effort goes into developing the secret sauce of business rules than maintaining a specific physical machine to sustain important workloads.

What’s the business value of SDN?

Lyons: SDN brings rigor to the living governance of infrastructure — auto-provision network elements only when you need them, so you have the tightest running ship from a business perspective. Historically, we relied on engineers for each change to physical infrastructure, and now there’s more speed and accuracy with software-driven changes.

In hybrid cloud environments, SDN applies when you need to quickly stand up virtual machines or add virtual containers for workloads, say at the end of each month, then tear them down when you’re done with them. SDN matches the software-defined compute because you can control the network at an API level and ideally bill clients just for what they consume, which is attractive for service providers. Where cloud providers offer standardized APIs, you can take advantage of mulicloud through service providers.

Overall, SDN offers amazing advantages but you can create unnecessary complications if you don’t have strong IT governance in place. For example, you need to make sure you don’t create trombone flows and drag traffic back unnecessarily across the network. Governance helps you simplify operations, which makes troubleshooting easier.

What steps can clients take now for successful cloud migration?

Lyons: While it’s tempting to think that a cloud migration allows you to walk away from the old ways of doing things, this approach is risky. To help ensure success, a good governance model and affinity maps are critical to understand how current and future workloads affect your business so you can navigate risks. Vendors use specialized tools to analyze the applications and create affinity studies so you can make careful decisions about what’s suitable to go to the cloud. It’s best if this is done in a technology-agnostic way so you’re not locked into a platform and can apply IT advances without impacting workloads.

Where does network capacity fit into this?

Lyons: Most clients have built key applications over time assuming that their networks will be always on and have sufficient capacity. What’s often missed is the implicit reliance on network latency and the impact this has on application performance. Many cloud migrations run into trouble when some application components are moved to the cloud and others are kept on premises, then discover network latency introduced between components affects performance. Effective IT governance helps you model this type of scenario and avoid problems.

Mike Lyons is an executive IT architect within IBM Services Network Domain of Technology Innovation and Automation group, which is responsible for developing service roadmaps, reference architectures and thinking about what’s next in the networking space. You can often find him on calls with clients around the world providing technical leadership and resolving issues with critical business impact. Follow him on Twitter @infoinmotion