fivelayers.net

Traditional network provisioning and change management process that are tied to post-sales day-1 and day-2 activities isn’t optimal anymore. They are prone to manual errors, difficult to track changes and involves considerable deployment risk. Provisioning issues can creep in when the design decisions are not correctly implemented and tested. Change management in many scenarios are operational nightmares. Additionally, for large and complex network configurations, the procedural automation scripts are not well suited as they are prone to errors and talent is scarce.

But things doesn’t have to be this way any more. To address these issues Cisco has developed a solution called SaC (Services-as-Code) that automates the whole network provisioning and change management process. SaC relies on the principles of IaC (Infrastructure-as-Code) that is widely popular in the hyperscaler (AWS, Azure, etc,) world for provisioning cloud services. SaC leverages declarative automation that abstracts away the implementation details and makes it simple to perform provisioning and change management activities. Recent research has shown that customers who adopt SaC are benefiting from 65% cost savings and 80%-time savings per change. 

The SaC System

The SaC automation system attaches to the controller systems like ACI and SDWAN and comprises of controller data model that holds the configuration data, a tool chain that includes git , terraform, ansible, automated testing and so on. These various components that make up the SaC system enables the DevOps motion and allows network operators (NetDevOps) to leverage a declarative or intent based approach to perform provisioning, configuration , deployment and testing activities on the network.

The magical thing about the declarative approach is that it abstracts away the complexity of writing procedural scripts for provisioning and change management. All network changes are made inside a simple YAML (Yet Another Markup Language) file data structure that holds the controller configuration data.

Unique Value Proposition

Some people might perceive that SaC solution is a bunch of open source tools stitched together with automation scripts, but it is only the contextual layer of the whole solution. The core layer of the SaC solution consists of several capabilities. Firstly, the data structure performs automated verifications before pushing configuration onto the controller (e.g., ACI) . Secondly the system validates the configuration against the best practices that are created by Cisco engineers over the years. This is crucial because the system prevents mis-configuration if it finds any deviation from the best practices. Finally, the solution has Integrated testing before pushing the configuration into controller and makes sure that the controller fabric is configured as intended. Customers can choose to use their own DevOps process and tool chain and layer-in on capabilities that are unique to SaC.

The Generative AI Part

With the recent addition of GenAI as part of the SaC system, the declarative approach is somewhat on steroids. What i mean by this is that the network operators will leverage natural language Interface like ChatGPT to perform provisioning and change management activities. The feature is called Cisco AI assistant that serves as a companion to the Network operator. At the outset, there are 3 areas that is being considered where GenAI can become a boon to the network operations team. They are a. Gen AI as the knowledge base (e.g., can you show me the config of the prod tenant?), b. Implementation assistance (e.g., can you write me a merge request description for the above changes? ), c. troubleshooting assistance (fault analysis). Please note that these features are currently under development.

The Partner Play

Starting April 2024 , eligible Cisco partners can sell and deliver SaC solution and service to their customers. SaC is an add-on component to Cisco’s new premium support offer called Partner Advanced Support(PAS) . SaC is sold as a subscription service that has embedded 24/7 technical support from Cisco, access to new IaC data models, ServiceNow and Thousand Eyes connectors and standard test cases.

The delivery of the SaC service requires Cisco partner to incubate the practice or in other words augment their architecture practice with automation capabilities from SaC. A Cisco and Partner co-delivered model can be created wherein partner owns the responsibility to deliver certain stages of the service and Cisco owns the rest. A partner branded, partner delivered model is the end goal in this co-delivered model.

Cisco Partner Success team supports partners in the SaC journey across the following stages from business alignment until the service launch.

Partner Value Added Services

Cisco partners can extend the value of SaC by adding in additional capabilities that are not inherently present within the solution.

Professional Services

Based on the customer requirement, partners can develop their own custom automation scripts. Additional opportunities include, customer tool chain configuration services, testing and validation services of control and data plane before the change in lab environment, DevOps coaching, etc.,

3rd Party Integrations – Expanding the Connector Eco-System

Partners have a great opportunity to expand the connector ecosystem and monetize on them. While SaC comes with connector for ServiceNow (SNOW), partners can build their own ITSM connectors for Jira, Remedy and so on. Another example where when customer wants to develop ‘auto-approve’ automation process for light-weight network change. Partner can build this level of automation

These are just few examples where partners can value add and monetize with their customers.

Additional Resources

Cisco Live Session – Unleashing the Power of Services as Code for your automation journey – Markus Harbeck – Technical Deepdive

Credits : Thanks to the following folks from Cisco product management and delivery team who have been the inspiration and source of the above knowledge through partner discussions and the great content they have developed

Phil Maynard, Rafael Muller, Amrit Wadhwa, Greg Leider