OpenFlow Switch Flow Entry Statistics Collection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In OpenFlow networks, the separation of data forwarding and control planes leads to incomplete data flow statistics due to each data flow being forwarded after matching between multiple flow entries, making it impossible to obtain complete information about a single data flow from a flow table, thus hindering effective network management.
Innovation Solution
A method and apparatus for data flow statistics collection that involves matching fields in a data packet with flow entries across multiple tables, recording identification information in a global statistical identifier, and performing statistics collection on the to-be-counted flow object, ensuring that each piece of identification information is recorded in the last flow entry corresponding to the object, allowing for comprehensive statistics collection and analysis.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If multiple flow tables are introduced to implement level-by-level query and forwarding, then flow table compression is achieved, but complete information of a single data flow cannot be obtained from a flow table
Solution Approach 1:
The patent segments the flow table into multiple flow tables (first flow table, second flow table, etc.) organized in a hierarchical structure. Each flow table contains flow entries that match specific parts of data flow characteristics. This segmentation enables level-by-level querying while maintaining the ability to trace complete data flow information through the hierarchical structure.
Solution Approach 2:
The patent introduces a global statistical identifier as an intermediary element that connects flow entries across multiple flow tables. This identifier is recorded in each flow entry and serves as a link to aggregate statistics from different flow tables, enabling complete data flow statistics to be reconstructed by following the identifier chain through the hierarchical flow table structure.
2Adaptability or versatility
If each data flow is forwarded after matching between multiple flow entries, then flow forwarding flexibility is improved, but statistical data of the data flow cannot be acquired from a counter field in a flow entry
Solution Approach 1:
The patent performs preliminary action by recording the global statistical identifier in each flow entry during the flow entry creation phase. This identifier is pre-configured to link flow entries across multiple flow tables that will match different parts of the same data flow. By establishing these links in advance, the system enables accurate statistics aggregation without requiring complex real-time correlation when data flows are being forwarded.
3Ease of operation
If a single counter field in each flow entry is used, then statistics collection is simple, but complete information of a single data flow cannot be obtained when multiple flow tables are used
Solution Approach 1:
The patent merges statistics from multiple flow entries across different flow tables by using the global statistical identifier as a common key. The controller aggregates counter values from flow entries that share the same global statistical identifier, effectively combining statistics from segmented flow tables into complete data flow statistics. This merging approach maintains the simplicity of individual counter fields while achieving complete statistics through centralized aggregation.
Data Source
AI summary
A data flow statistics collection method, a switch, and a controller. The data flow statistics collection method includes recording first identification information and second identification information in the first flow entry if a part of fields in a data packet match a flow entry and the flow entry is the last flow entry corresponding to the to-be-counted flow object, where the first identification information identifies that the first part of fields match the first flow entry and the second identification information identifies that at least one second part of fields in the data packet respectively match a second flow entry in at least one second flow table. The data flow statistics collection method also includes performing, in the first flow entry, statistics collection on the to-be-counted flow object, and recording statistical information of the to-be-counted flow object. In this way, statistics collection on the to-be-counted flow object is implemented.