EMA: Network Management Tools Must Evolve for SDN

Shamus McGillicuddy, Senior Analyst for Enterprise Management Associates (EMA), recently published a comprehensive research study on “Managing Tomorrow’s Networks: The Impacts of SDN and Network Virtualization on Network Management.” Based on a survey of early adopters, the research explored how prepared enterprises and service providers are to manage SDN, NFV, and network virtualization. The short conclusion is that they are not very prepared, and the path forward is not very clear.

In case you don’t want to purchase and wade through the 95-page report, we’ve summarized a few other conclusions from the service provider section of the study here. It is based on the responses of 76 early adopters of SDN within communication service providers (CSPs) in North America, Europe, and the Asia-Pacific region. The majority of these CSPs have some production of SDN deployment completed. (EMA also prepared a special, six-page report on the service provider results for Packet Design, which you may download here.)

Overall, EMA’s conclusion is that “network management tools must evolve for SDN,” which is something Packet Design has been saying for a long time.

McGillicuddy writes:

“From planning and engineering to availability monitoring and performance monitoring to troubleshooting, SDN adopters have clearly indicated a need for new features and functions that are unique to software-defined networks. Planning and engineering tools need to understand overall network state for better analysis and simulation, and SDN adopters need to acquire capacity planning features that are tuned to SDN flows. Availability monitoring systems need to collect new SDN metrics, especially in virtual environments where a new layer of infrastructure adds to overall complexity. Performance monitoring tools need to be responsive to the dynamic nature of SDN, and they need to provide visualization of these dynamic paths. Troubleshooting tools need to analyze and report on events that are unique to SDN.”

With this context in mind, here are the main findings of the survey:

• The majority of network service providers have determined that their existing activation and provisioning, operations support and readiness, and service assurance tools do not fully support SDN and NFV.
• Ironically, the further CSPs go in deploying SDN, the less confident they become in their existing tools. Of CSPs that have not yet deployed SDN, 52% believe their existing performance management systems will fully support SDN versus only 31% with advanced deployments. Similarly, only 10% of the first group believes separate SDN management tools will be needed compared with 28% of CSPs with advanced deployments.
• Both enterprises and service providers are divided on how to address these management system shortcomings. Many are modifying existing tools, and many more are acquiring new tools.

Because the move to SDN and NFV is a business decision as well as a technology one for CSPs, EMA asked about the single biggest benefit they hope to realize through their SDN and NFV efforts. Here are the top three:

1. Improved productivity of end users (28%)
2. Revenue growth (25%)
3. Operational trust (17%): This required a long explanation in the report, but it is basically a military term referring to being able to trust each individual component in support of a shared goal. SDN, when managed correctly, should facilitate this: “Service providers are making a calculated bet that SDN and NFV will reduce the need for top-down management of network infrastructure. It will allow individual units within the organization to create, manage, and support services independently. SDN will also enable business and infrastructure leaders to set high-level goals for the service provider organization that individual units can work independently to achieve.”

What then do CSPs need to make all this work and realize these benefits? McGillicuddy writes: “What has become clear to early adopters of SDN is that with all this dynamism, network operators need their networks to push telemetry to management systems. These management systems need granular visibility into changes occurring in the network. They must be able to model and predict network behavior based on this telemetry. And they must link this predictive analysis to business policy, for instance, by giving CSPs the ability to predict how programmatic changes to the network will impact SLAs.”

With this in mind, here are the top priorities of CSPs in the three general categories of network management tools that EMA identified in the report:

Service Activation and Provisioning Management

• 46% want to be able to specify polices based on advanced modeling of network configuration and operational state
• 46% want to be able to activate services based on it

Operations Support and Readiness

• 47% want the ability to implement pre-approved, software-based changes to networks and services
• 37% want a real-time topology view and inventory mapping system
• 36% need logically isolated test environments for change management

Service Assurance

• 41% want the ability to perform end-to-end service quality management across physical and virtual resources
• 38% need increased automation of network optimization
• 37% want predictive, service-centric traffic modeling

Read more results in EMA’s Packet Design-specific report – “Real-Time Network Telemetry Is Essential to SDN Management and Orchestration” – including their perspective on our SDN Management and Orchestration platform.