Intel White Paper Closed-Loop Automation - Telemetry-Aware Scheduler for Service Repair and Platform Recovery

White Paper Intel Corporation

Closed Loop Automation - Telemetry Aware Scheduler for Service Healing and Platform Resilience

Authors 1 Introduction John Browne Closed loop automation is the process of continually monitoring real-time network

conditions, workload requirements, and resource capabilities and availability to

Emma Collins

determine the optimum workload placement for faster and more efficient delivery of Krzysztof Kepka services. Service Providers are adopting Closed Loop Automation as a strategic priority to address their business objectives [1.]. One such objective is to ensure customer Service

Sunku Ranganath

Level Agreements (SLAs) are met to provide an optimum Quality of Experience (Qo E), Jabir Kanhira Kadavathu which is delivered according to operator defined policies. Network Services rely on the infrastructure they run on and so this is a crucial place to start with when trying to ensure

Swati Sehgal

an optimal Qo E.

Killian Muldoon

This document provides an overview of a platform resiliency prototype that showcases Michal Kobylinski the integration of key Closed Loop Automation components, leveraging a mix of solutions provided by Intel and the open source community. Traditional Telecom Services Assurance is an Operations Support System (OSS) function carried out offline [1.] without automated processes and methods for self-optimization of the network. This paper outlines ‘near real-time’ automation at a granular platform level. This allows an orchestration system to detect and respond to an issue before a customer is impacted, including integration with Management and Network Orchestration (MANO), using platform telemetry, and analytics. Closed loop solutions integrate components that include platform, telemetry, analytics processing, MANO and polices to create automated processes. The Intel platform has a wealth of features and resources that provide a rich set of data and control points that can be used for configuration, reporting, monitoring and managing workloads in a network infrastructure. This data can be used as part of a Closed Loop Solution to provide granular insights into the behavior of the system. When these metrics provide actionable alerts, the configuration controls can be used to provide platform resiliency support and ensure minimal unplanned operational downtime. By collecting Intel® Architecture specific feature telemetry, this prototype will provide host insights to a new Kubernetes* (K8s*) scheduling extension, called Telemetry Aware Scheduling (TAS), to trigger corrective healing actions on a workload and influence workload placement decisions, depending on the reliability of the underlying infrastructure. By gaining these insights into the health of the platform, the services they are running on, and having the control to be able to remediate against faults and errors in an automated way, this prototype provides the methodology to reduce risk of unplanned outages and enable scheduled downtime for maintenance, therefore maximizing service availability. Along with the video showing this use case in action included in the release, this document is part of the Network Transformation Experience Kit, which is available at: https://networkbuilders.intel.com/network-technologies/network-transformation-exp- kits