Parallel VNF Processing in Service Function Chains
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Virtualized service function chains (SFCs) experience reduced throughput and increased latency due to the limitations of network function virtualization (NFV) and software-defined networking (SDN), which can lead to longer SFC lengths and management complexities, making it challenging to maintain acceptable latency for latency-sensitive applications.
Innovation Solution
A hybrid packet processing system that dynamically distributes data packets to virtual network functions (VNFs) in parallel and intelligently merges the output, ensuring the preservation of sequential processing semantics, while leveraging a lightweight mirror and merge function to reduce latency and improve throughput.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If network functions are virtualized and processed sequentially through SFCs, then service provisioning flexibility is improved, but processing latency increases and throughput decreases
Solution Approach 1:
The patent segments the sequential SFC processing into parallel processing paths by identifying independent VNFs that can operate simultaneously. The mirror function creates duplicate packet copies that are distributed to multiple VNFs in parallel, while the merge function recombines results. This segmentation transforms the linear processing chain into a parallel architecture, reducing overall processing latency while maintaining service flexibility.
Solution Approach 2:
The patent introduces a temporal dimension to packet processing by implementing out-of-order delivery and buffering mechanisms. Packets are processed in parallel across multiple VNFs and may arrive at the merge function in different orders. The system buffers these packets and reassembles them in the correct sequence, adding a time-management dimension that enables parallel processing without sacrificing sequential correctness.
2Adaptability or versatility
If more VNFs are added to SFCs to enhance service capabilities, then service functionality is improved, but SFC length increases and processing efficiency deteriorates
Solution Approach 1:
The patent merges the processing capabilities of multiple VNFs into parallel execution paths. Instead of chaining VNFs sequentially where each packet must pass through every VNF in order, the system identifies VNFs with no data dependencies and merges them into parallel groups. This merging reduces the effective processing path length and improves throughput while maintaining the combined functionality of all VNFs.
Solution Approach 2:
The mirror function creates copies of packets that are distributed to multiple VNFs simultaneously. This copying mechanism enables multiple VNFs to process the same packet data in parallel without waiting for sequential processing, thereby maintaining service functionality while dramatically improving processing efficiency and reducing latency.
3Speed
If parallel processing is implemented for VNFs, then processing speed is improved, but system complexity increases due to merge operations and dependency management
Solution Approach 1:
The patent introduces mirror and merge functions as intermediary components that manage the complexity of parallel processing. These intermediaries handle packet distribution, copying, buffering, and reassembly, isolating the complexity from the core VNF processing logic. The mirror function acts as a mediator that creates parallel processing opportunities, while the merge function mediates the recombination of parallel results, maintaining simplicity in the VNF implementations themselves.
Solution Approach 2:
The system implements self-service mechanisms where VNFs include built-in dependency information that automatically enables the control plane to identify parallelization opportunities. VNFs publish their own data dependency characteristics, allowing the system to automatically generate mirror and merge operations without manual configuration. This self-service approach reduces operational complexity while enabling parallel processing.
Data Source
AI summary
The concepts and technologies disclosed herein are directed to parallelism for virtual network functions (“VNFs”) in service function chains (“SFCs”). According to one aspect, a packet processing system can receive instructions to process, in parallel, at least a portion of a plurality of data packets associated with an SFC including a plurality of VNFs. The system can create a copy of at least the portion of the data packets. The system can send the copy of at least the portion of the data packets to at least two VNFs. The at least two VNFs can process, in parallel, the copy of at least the portion of the data packets. The system can receive, from the at least two VNFs, processed packets including the copy of at least the portion of the data packets and processed, in parallel, by the at least two VNFs. The system can combine the processed packets.