Despite countless tools and technologies, network automation takes time, patience, and is difficult for enterprises that have grown organically over the last twenty years.
There is more time spent on tools analysis than on truly understanding the workflows and data required to drive network automation. Moreover, there is a lack of focus on the integrations required to truly create a holistic network automation solution. This must change.
In order to streamline adoption and ensure greater success with network automation, one must ask, “How can organizations truly start to embrace network automation?”
The answer starts with data.
Data as the foundation
Network data–provided it’s clean, quality data–is the true foundation of network automation.
While it may sound simple, the process of data curation and management proves to be a far more difficult task. For many, it is one of the most arduous tasks due to the fact that today's IT infrastructure is dictated by a combination of spreadsheets and numerous data sources owned by different teams and stakeholders.
However, a data-first approach is necessary for successful network automation. It attacks the problem head-on and provides the path for long-term success.
Accurate network data is needed to define intent, allowing users to understand and drive the true desired configurations and policies. Defining intent and using the intent data to automate the network is how network operators can ensure the approved standards and designs are always used. For the past 25 years, the only truth was the configurations running on the network devices. There was no source of truth decoupled from the infrastructure to know what should be running on those devices. That has to change for successful network automation.
In an ever-changing multi-vendor, multi-domain, and multi-cloud world, network automation needs a consistent and holistic approach. As automation increases operational speed, the gap in trusted sources of data and documentation has become a larger liability. Having a data-first approach brings consistency from day one.
Data and a Source of Truth
A fully integrated, centralized Source of Truth (SoT) is a critical foundational element of modern network automation architectures.
Without a centrally accessible, programmable, and authoritative source of network information, network teams are operating in the dark, relying on multiple, potentially inaccurate sources with redundant and stale data. These teams have nowhere to define the intended state of the network nor a trusted source with which to verify the current state for compliance against that intent.
Having clean and quality data inside the Source of Truth results in trusted data being deployed by an automation platform and onto the network. Only when data is normalized and centralized with a Source of Truth will there be continued firefighting and a lack of trust with the network automation tooling.
With this approach, the centralized reference point defines the intended state of the network. The intended state is documented and stored in multi-vendor data models spanning both desired configuration and the operational state. A Source of Truth is the foundation for a data-first approach to network automation, making it sustainable and creating a solid plan for long-term success.
Source of Truth Extensibility
What should organizations be thinking about when it comes to SoT? Extensibility, for one. The SoT must become an extension of the network, providing extensibility to cater to both traditional software-defined and cloud networks. It should be possible to model any network in the Source of Truth. Further, integrating with NetDevOps workflows becomes more important by the day as DevOps principles, tools, and processes continue to gain traction in networking. This includes ensuring there are integrations with Git, which is often the starting point for defining data in YAML files, but also APIs and event triggers when performing CRUD operations on the data.
Open Source Deployment
While CMDBs exist, there really aren’t commercial solutions focused on being a Network Source of Truth. Open Source has been the primary means to deploy a source of truth because of the need to provide abstract and multi-vendor data models. This is a good thing as it allows for the highest level of interoperability between devices from all vendors. Of course, with open source also comes industry-wide collaboration, community, flexibility, and control.
Programmatic Sources of Truth
Independent of how it’s done, programmatic Sources of Truth are an absolute requirement for network automation. While anything is possible, it is not ideal to programmatically fetch, update, and manage data in text documents or spreadsheets. Spreadsheets must be eliminated while there is also consolidation of where data is stored.
Simply put, with a Source of Truth, it becomes possible to document the intended network state and realize the full potential of network automation.
The future is data-driven network automation
Network automation is a journey, not a destination, just like networking.
Is an organization ever done doing networking? Devices change, requirements evolve, new technologies emerge, priorities change, and the list goes on. The same goes for network automation. The best way to design for change and plan for the unknown is to define the intended state.
Data-driven network automation and the concept of the intended state will only grow in the coming years. Without a solid Source of Truth strategy, the network automation deployed today will turn into the network management of yesterday.
We have seen the need for a Source of Truth, and now it is time to embrace it.
Jason Edelman is the founder and CTO of Network to Code.