NRF Overload Protection via Exponential Backoff Probing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In 5G mobile communication networks, Network Repository Functions (NRFs) often become overloaded, leading to inefficient operation, as continuous probing by Network Functions (NFs) exacerbates the load and prevents NRFs from reducing their workload, causing cycles of high load and inefficiency.
Innovation Solution
Implementing overload protection by providing NFs with load metrics and validity time attributes to manage NRF interactions, using a backoff algorithm to determine when to resume interfacing with overloaded NRFs, and dynamically distributing load across NRFs to prevent overloading.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If NFs continuously probe NRFs to maintain service discovery functionality, then service discovery reliability is improved, but NRF load increases causing overload
Solution Approach 1:
The system implements periodic probing with exponential backoff intervals instead of continuous probing. When an NRF is detected as overloaded, the probing interval is extended exponentially, reducing the frequency of probe messages sent to that NRF while maintaining eventual detection capability. This transforms continuous harmful action into periodic controlled action.
Solution Approach 2:
The system uses load metric feedback from NRFs to dynamically adjust probing behavior. NFs receive load metrics from NRFs and use this feedback to determine whether to continue normal probing or switch to extended intervals. This closed-loop feedback mechanism allows the system to adapt probing intensity based on actual NRF conditions, preventing overload while maintaining service discovery reliability.
2Speed
If NFs send frequent probe messages to detect NRF status, then overload detection speed is improved, but network signaling overhead increases
Solution Approach 1:
The system dynamically adjusts the probing interval based on detected NRF conditions. When an NRF is healthy, probing occurs at normal intervals. When overload is detected, the interval extends exponentially. This dynamic adjustment optimizes the balance between detection speed and signaling overhead, sending more messages when needed for fast detection and fewer messages when the system is stable.
Solution Approach 2:
The system changes the temporal parameter (probing interval) based on system state. The interval transitions from a fixed normal value to an extended value calculated using exponential backoff. This parameter change allows the system to reduce signaling overhead during stable periods while maintaining the capability for rapid detection when conditions change.
3Speed
If NFs immediately re-register with recovered NRFs, then service restoration speed is improved, but NRF load spikes causing re-overload
Solution Approach 1:
The system performs preliminary assessment of NRF recovery status before allowing full re-registration. Instead of immediately allowing all NFs to re-register with a recovered NRF, the system first verifies the NRF can handle the load and gradually restores services. This preliminary action prevents sudden load spikes that would cause re-overload.
Solution Approach 2:
The system uses periodic probing with exponential backoff intervals to gradually restore service to recovered NRFs. Rather than immediate full-capacity re-registration, NFs resume probing at extended intervals that gradually decrease as the NRF stabilizes. This periodic approach with controlled intensity prevents load spikes while maintaining service restoration capability.
Data Source
AI summary
Providing overload protection may include receiving, from a first NRF of two or more NRFs, a load metric and comparing the load metric to a threshold. Additionally, when the load metric is above the threshold, providing NRF overload protection may include sending a Network Function (NF) Discovery (NFD) message to a second NRF instead of the first NRF and sending a heartbeat signal to the first NRF.