To understand the change in thinking represented by modern networking, I’ve been considering a hypothetical network engineer. A kind of stimulus driven, lizard-brain creature, who is a product of their environment and may fear the changes occurring. This is not a single person, but a collective of network engineers who are in various stages of emergence into a software driven world…
A network engineer has trained for years and built their specialist knowledge to understand her networks inside and out. A network engineer creates in her mind a picture of the network and how she expects interactions with the network to proceed. That picture contains the general connectivity methods and paths through the network, individual device roles, including how they are managed and monitored and how they are configured to allow certain behaviour. Individual device config is curated to fit her mental map. It is the ability of the network engineer to keep a detailed picture in her mind that allows her to spot misconfigurations in planned changes or troubleshoot a complex issue with high impact. This, then, is what the network engineer sees as their advantage: her comprehension of the ways in which you can interact with a network without impact.
The network engineer, being socially responsible, wishes to minimise the amount of impact felt when problems happen. If the network deviates from his mental map, he cannot be sure that the behaviour is expected. So to keep sane, the network engineer also develops policies that ensure the picture does not change too quickly. Many networks operate on the knowledge of an individual who intimately understands the workings of the network. If new devices, paths or connections are proposed, he assesses whether that deviates significantly from his own mental map. If so, you may receive a polite rejection of your suggestion on the grounds that it is not technically feasible. In establishing policies that restrict change, the network engineer understands that change is possible, but too much change and the mental map will not be updated quick enough to maintain his advantage.
Hence, time is the real problem, time to understand the change, simulate its impact and update the mental map. The network engineer is a human! not a machine!
This, then, is the shift in thinking that can drive change in the network. Reduce the time to make changes to the existing network. What can be done speed up the deployment of network changes without introducing risk? What new systems and processes can be put in place to facilitate the accelerated rate of change?
The network engineer must let go of the mental map and use a system that can define the network as a model.
Data models for distributed systems are not new, but applying them to networks is only beginning to emerge into mainstream network systems. This requires a change in thinking for individual network engineers. Under the new software driven world, changes to the network should be made according to the model. A system that allows a model to be created and configuration defined, does not require the network engineer to retain a detailed mental map; the system does that for them. Changes can then be implemented by users through the system and processes without the need for hand written implementation plans. The network engineer is then free to focus on enabling new network types and enhancing the model.
Operation model transformations such as this can be difficult to do. Diaxion specialise in helping organisations through these changes. Contact us for some advise on your situation.