Hitless Data Pipeline Reconfiguration via Staged Image Updates
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The existing data processing pipelines in hardware are fixed during manufacturing, making it impossible to add, delete, or modify functionalities after deployment, limiting flexibility and adaptability to changing network requirements.
Innovation Solution
A method and apparatus for a network element that allows hitless reconfiguration of the data processing pipeline by preparing a new image, putting the pipeline into a non-forwarding mode, writing the new image into memory, and then switching back to forwarding mode, enabling the addition, deletion, or modification of functionalities without downtime.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a fixed ASIC data processing pipeline is used, then manufacturing reliability is improved, but adaptability deteriorates
Solution Approach 1:
The data processing pipeline is segmented into multiple functional stages (ingress interface, parser, classifier, action profiler, egress interface) that can be independently configured and programmed. Each stage can be modified without affecting the entire pipeline, enabling adaptability while maintaining manufacturing reliability of the overall ASIC structure.
Solution Approach 2:
The pipeline transitions from a static fixed configuration to a dynamic reconfigurable system through the use of state machine controllers and programmable logic. The pipeline can be dynamically reprogrammed via control plane software to adapt to changing network requirements while maintaining hardware-based processing reliability.
2Adaptability or versatility
If the data processing pipeline is reconfigured after deployment, then adaptability is improved, but service continuity deteriorates
Solution Approach 1:
Configuration changes are prepared and validated in advance in the control plane before being applied to the data plane. The control plane software compiles, validates, and stages configuration changes so that when they are applied to the pipeline, service continuity is maintained without interruptions or churn.
Solution Approach 2:
The control plane acts as an intermediary between the network management system and the data plane pipeline. It mediates configuration changes by validating, compiling, and staging them before application, ensuring that adaptability is achieved without disrupting service continuity in the production network.
3Manufacturing precision
If the pipeline is put into non-forwarding mode for reconfiguration, then configuration accuracy is improved, but productivity deteriorates
Solution Approach 1:
Configuration changes are staged and validated in the control plane before being applied to the data plane. This preliminary preparation ensures configuration accuracy by catching errors before they affect production traffic, while minimizing the time the pipeline needs to be taken offline.
Solution Approach 2:
The system quickly transitions through the non-forwarding state by using efficient state machine controllers and staged configuration application. The pipeline is taken out of service mode only for the minimum necessary time to apply configuration changes, then rapidly restored to forwarding mode to minimize productivity loss.
Data Source
AI summary
A method and apparatus of a network element that processes data by a network element with a data processing pipeline is described. In an exemplary embodiment, the network element prepares a new image for the data processing pipeline of the network element, where the data processing pipeline processes network data received by the network element and the new image modifies one or more functionalities of the data processing pipeline. In addition, the network element puts the data processing pipeline into a non-forwarding mode. Furthermore, the network element writes the new image into memory of the data processing pipeline. The network element additionally puts the data processing pipeline into a forwarding mode, where the data processing pipeline processes network data received by the network element using the new image.