As an analyst who focuses on network management research, I am particularly intrigued by software-defined networking (SDN). As SDN architectures are deployed in data centers, local area networks and WANs, network management practices will have to evolve.
For instance, SDN may make it easier for cloud administrators to provision network services and connectivity for a new application, but how do you ensure that your new programmable network remains compliant with configuration controls and policies? Is your performance management tool able to model and monitor an SDN controller? If you have traditionally relied about appliance-based load balancers and firewalls in your data center, how do you monitor and manage those network functions when they become virtualized services that are as mobile and dynamic as the workloads they serve?
SDN has always promised to make networking easier, but there have been very few discussions around how networking professionals will manage and operationalize these technologies. After all, making a technology easier to use doesn’t automatically make it well-managed. I have talked to several network management software vendors who are adapting their products to SDN, but its a difficult landscape to navigate. There are so many different architectures — hardware-based SDN underlays, hypervisor-based network virtualization overlays, open source controllers.
On May 13, the Open Networking User Group (ONUG) will hold its spring meeting. I am hoping that many of the sessions will address the management questions I have.
The event’s agenda includes presentations from various working groups that are exploring the management questions of SDN, including the Common Management Tools Working Group.
There is also an NSV Working Group presentation. NSV (Network Service Virtualization) is related to the Network Functions Virtualization architecture being specified by the service provider industry. NSV involves the transformation of network functions (load balancers, firewalls, WAN optimization controllers) from appliances to software. This change isn’t just a shift from hardware to software. It also involves the transformation of these functions into elastic services that can be deployed rapidly and flexibly. NSV offers a great deal of agility, but I’m still wary of the management and governance of these services. I’m hoping that the ONUG folks demonstrate some progress in defining how the industry should proceed.