Network Node Predicts Service Instances to Reduce Cloud Latency
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current serverless computing and virtual machine approaches in cloud infrastructure face high latency and inability to maintain persistent connections, limiting their effectiveness in communications networks due to the need for constant resource allocation and reinitialization.
Innovation Solution
A network node predicts and preloads service instances and network connections based on historical data using reinforcement learning, ensuring instances are readily available and resources are optimized, allowing for persistent connections and reduced latency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of energy
If serverless computing is used to deploy software in cloud environment, then resource allocation cost is reduced and resources are allocated on demand, but latency increases and persistent connections cannot be maintained
Solution Approach 1:
The patent applies preliminary action by predicting future service requests using reinforcement learning and pre-loading service functions into memory before they are actually needed. This allows the system to maintain low latency and persistent connections while still using serverless computing's cost-effective resource allocation. The service function is prepared in advance based on predicted demand patterns.
2Loss of energy
If serverless computing is used with on-demand resource allocation, then cost efficiency is improved, but the ability to maintain persistent connections is lost
Solution Approach 1:
The system predicts future service requests and pre-loads service functions into memory, maintaining connection state information ready for immediate use. This preliminary preparation enables persistent connections to be maintained while resources are allocated on-demand, resolving the contradiction between cost efficiency and connection stability.
Solution Approach 2:
The patent employs reinforcement learning to continuously monitor service usage patterns and adjust resource allocation decisions accordingly. This feedback mechanism allows the system to learn from actual usage behavior and optimize both cost efficiency and connection persistence by adapting resource management strategies over time.
3Reliability
If resources are allocated constantly to ensure availability, then service availability is improved, but resource waste increases
Solution Approach 1:
Instead of allocating resources constantly, the system uses reinforcement learning to predict when services will be needed and pre-loads only those specific service functions into memory. This targeted preliminary action ensures high service availability for predicted requests while avoiding resource waste by not maintaining all possible services in memory simultaneously.
Solution Approach 2:
The patent dynamically changes the state of service functions between loaded and unloaded based on predicted demand. The system adjusts resource allocation parameters in real-time, loading service functions into memory when predicted to be needed and unloading them when not expected, thereby optimizing both availability and resource efficiency.
Data Source
AI summary
A method performed by a network node for handling one or more upcoming requests for one or more services from a communication device in a communications network using a cloud infrastructure is provided. The network node obtains (301) a prediction of an amount of instances needed for the one or more upcoming requests for the one or more services, based on historical data of previous requests for services, and obtains (304) network connections needed for the one or more upcoming requests for the one or more services, taking the prediction into account. Furthermore, the network node decides (307) to instantiate a service with the obtained network connections and/or the amount of instances as predicted.