The dark NOC is the ultimate goal of network operations automation. We have been automating numerous aspects of the work of network operations engineers, but in recent years, the ultimate goal of a “Dark NOC” has started to seem like an attainable goal. The advent of advanced machine learning algorithms, growing promises of autonomy in areas like mobility, and other automation developments have made it look like a goal that should be within reach soon. So when can we expect to turn off the light in the NOC?
The “Dark NOC” is of course not a goal in itself and there are multiple reasons why it is so desirable to automate as much as possible of the work of a communication service provider (CSP). A number of the driving forces for that pursuit:
- Connectivity becoming a commodity is tightening the margins and requires reducing operation costs.
- The growing complexity of the networks due to new technologies like 5G
- Desire for end-to-end insights that require cross-domain analysis
- The continuing growth of the amount of data generated for monitoring purposes.
- Ever faster changing of the network due to the roll-out of new equipment, and features.
What we see now is a bit of a fragmented landscape of SON solutions from the equipment vendors and third-party vendors. These SON solutions usually don’t span multiple domains and are not meant to cope with all the possible disturbances in the network.
Even getting all the data together for structural end-to-end monitoring capabilities is a challenge. This is mostly caused by the number of different systems involved and by the fact that multiple departments are often involved.
In most organizations, projects have been started for experimenting with machine learning, but not many of the resulting models have been integrated with the network operations processes.
When pursuing the goal of a “Dark NOC”, it is wise to take steps in the right order, or you risk losing a lot of time. For example, setting up even the most basic end-to-end monitoring without a proper mechanism for collecting the data from multiple domains is very time-consuming and prone to failure. Realizing that the data plays a central role in this journey makes us see that we can in fact take the analytics maturity model as a base for a roadmap.
At the end of this model, we can place our ambitions for a fully automated NOC. AI and machine learning will be a key ingredient here, but the steps before that are required to transform the organization to be able to feed and properly use those tools.
A complete “lights-out” NOC will probably take many years to arrive, as a lot of companies are still very much working themselves up the ladder of the analytics maturity model. There are plenty of advantage-gaining steps in that direction that can be taken right now. Our 1OPTIC solution provides solid ground for that by consolidating, combining, and standardizing all network data from different domains and equipment types.
Feel free to contact us to find out how we can help you!