It is no secret that I am interested in network management. Every time I start working in a large network, I get frustrated with the lack of tools to see what is happening in the network at a system level. Seeing the volume and location of link errors, duplex mismatches, and other fundamental errors provides an understanding of the network’s level of maintenance.
Similarly, Network Change and Configuration Management (NCCM) provides information about the consistency of the configurations. Are configuration standards being maintained? Where are the exceptions to the standard configurations? If the configs are inconsistent, I know that we can expect more problems due to inconsistencies than if they were consistent. It also tells me how much manual configuration and troubleshooting effort are required to run the network. The main problem is that current networking configuration is done on a per-box basis
One of the benefits of SDN is a logically centralized control system. While that doesn’t necessarily mean that all the consistency problems will disappear, it does suggest that making things different will require more work than making them the same. I wrote a blog at nojitter.com back in September 2013 titled Will SDN Be the Future of Network Change Management? The switch to SDN will enable us to make significant forward progress in network management, especially in the area of change and configuration management. Instead of working with the configurations of individual boxes, we’ll be working with network configurations. The key will be to determine the abstractions that enable us to work efficiently at a higher level. The details like duplex mismatch will be handled in a standard way within the SDN configuration and control system. At a higher level, I expect things like VLAN definitions to be replaced with SDN network slices, which will accomplish network traffic segregation.
SDN is giving us a chance to improve the way we manage and configure networks. And I think that’s pretty exciting.