TCAM Range Rule Testing via Interval Segmentation and Gray Coding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Hardware-based packet switching technologies, such as those using ternary content addressable memory (TCAM), face high hardware costs due to the need for a large number of memory entries to implement range rules effectively, which can bottleneck packet switching speed and increase network device costs.
Innovation Solution
The method involves defining non-overlapped intervals on a range field, assigning interval identifications (RIDs) to these intervals, encoding RIDs into TCAM entries, and using Gray coding to reduce the number of TCAM entries required for range rules, thereby reducing hardware costs and improving packet switching speed.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If hardware-based rule testing using TCAM is used to improve packet switching speed, then packet switching speed is improved, but hardware cost increases due to large number of memory entries required for range rules
Solution Approach 1:
The patent segments the range field into multiple non-overlapped intervals, each assigned a unique interval identification (RID). Instead of storing complete range rules in TCAM, the system stores only interval boundaries and their RIDs. This segmentation transforms a single large range rule into multiple smaller interval segments, dramatically reducing the number of TCAM entries needed while maintaining packet switching speed through efficient interval matching.
Solution Approach 2:
The patent introduces a new dimension by encoding interval identifications (RIDs) into the packet header alongside traditional match criteria. This additional dimensional information allows the system to perform range matching through a different mechanism - by matching RIDs rather than comparing full range values directly in TCAM. This dimensional transformation reduces the complexity of TCAM entries while preserving the speed benefits of hardware-based matching.
2Device complexity
If range rules are implemented in software to reduce hardware cost, then hardware cost is reduced, but packet switching speed decreases due to insufficiently fast software execution
Solution Approach 1:
The patent introduces an intermediary mechanism - the interval identification (RID) system - that bridges software and hardware approaches. RIDs serve as compact representations of range intervals that can be efficiently stored in hardware (TCAM or regular memory) while enabling fast matching operations. This intermediary allows the system to achieve hardware-speed performance without requiring expensive full-range TCAM entries, effectively mediating between the speed requirements of hardware and the cost constraints of software implementation.
3Reliability
If complete range rules are stored in TCAM to ensure accurate rule matching, then rule matching accuracy is maintained, but the number of TCAM entries increases leading to higher hardware cost
Solution Approach 1:
The patent creates a simplified copy of the range rule information by extracting and storing only the essential interval boundary points and their corresponding RIDs in TCAM, rather than storing complete range rules with all their detail information. This copying approach maintains rule matching accuracy because the interval boundaries define the complete ranges, while dramatically reducing TCAM entry size and quantity. The full range rule information can be reconstructed or retrieved from software storage when needed.
Data Source
AI summary
Aspects of the disclosure provide an improved method for processing a packet transmitted on a network. The method for processing the packet can include defining a set of non-overlapped intervals on a range field, defining one or more range rules that each range rule is in association with a subset of the non-overlapped intervals, identifying a non-overlapped interval, to which a parameter of the packet belongs, and executing an action on the packet based on a range rule associated with the non-overlapped interval to which the parameter belongs.