Just-enough-time Service Function Chain Resource Provisioning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional distributed routing protocols in multi-domain networks fail to consider service function availability, leading to unpredictable delays and high computation complexity in service function chaining, especially in resource-constrained environments like cloud and big data applications.
Innovation Solution
A vertex-centric distributed approach that orchestrates service function chaining by mapping service functions to physical nodes based on resource availability in the time dimension, allowing for pre-emptive instantiation of resources before packet arrival and efficient resource allocation and deallocation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional distributed routing protocols are used to compute network paths, then routing simplicity is maintained, but service function availability is not considered leading to unpredictable delays
Solution Approach 1:
The system performs preliminary actions by pre-computing service function chain paths and pre-instantiating virtual machine resources before actual service requests arrive. This advance preparation ensures that when packets need to be processed, the service functions are already available and resources are ready, eliminating unpredictable delays while maintaining reliability.
Solution Approach 2:
The system implements self-service through automated resource provisioning and service function chain computation. The network infrastructure automatically discovers available service functions, computes optimal chains, and provisions resources without manual intervention, ensuring service function availability is maintained while reducing delays through automated real-time adaptation.
2Ease of operation
If centralized orchestrator is used to maintain hierarchical topology for all domains, then service function chain mapping is simplified, but computation complexity increases significantly
Solution Approach 1:
The system segments the centralized orchestrator functionality into distributed components located at different network domains and service function nodes. Each segment handles local service function chain computation and resource provisioning independently, dividing the overall computation complexity into manageable pieces while maintaining ease of operation through coordinated segmentation.
Solution Approach 2:
The system transitions from a single-dimensional centralized control model to a multi-dimensional distributed architecture where service function chain computation occurs across multiple dimensions: local domain level, inter-domain level, and end-to-end service level. This dimensional transformation reduces computation complexity at any single point while maintaining overall mapping simplicity.
3Adaptability or versatility
If virtual machines are launched on-demand, then resource flexibility is improved, but setup time becomes highly variable and difficult to predict
Solution Approach 1:
The system applies preliminary action by pre-instantiating virtual machine resources in advance based on predicted service function chain requirements. Instead of launching VMs only when immediately needed, the system anticipates demand and prepares resources beforehand, maintaining resource flexibility through selective pre-provisioning while dramatically reducing and stabilizing setup times.
Solution Approach 2:
The system implements dynamics by making the resource provisioning system adaptive and flexible through real-time monitoring and dynamic adjustment. The system can dynamically scale resources up or down based on actual traffic patterns and service requirements, maintaining resource flexibility while reducing setup time variability through intelligent dynamic resource allocation rather than static on-demand provisioning.
Data Source
AI summary
A resource orchestration framework may include vertices representing physical network nodes and a resource orchestrator. The resource orchestrator may obtain a service function chain specifying, for multiple service functions, mappings of the service functions to respective physical nodes along a specified forwarding path, and a first starting time at which to instantiate resources for a first service function in the chain on the first mapped physical node. The resource orchestrator may instantiate the resources for the first service function on the first mapped physical node at the first starting time, prior to arrival of a first packet in a packet flow for the service function chain at the first mapped physical node and, subsequently, instantiate resources for a second service function in the chain on the second mapped physical node at a second starting time, prior to arrival of the first packet at the second mapped physical node.