Packet Registration System Using Hash Table Filtering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional methods for registering associations between inner and outer headers of encapsulated packets in a database are inefficient, leading to high processing loads and slow speed processing due to the need to access and retrieve all associations.
Innovation Solution
A registration system that includes an extracting unit for extracting inner and outer header information, a filter unit that uses a hash table to selectively pass only the first packet of a series of flow, and a registering unit that associates and registers this information in a database.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If all inner header and outer header associations are registered in the database, then complete packet analysis capability is achieved, but processing load increases and registration speed decreases
Solution Approach 1:
The patent segments the packet flow into first packets and subsequent packets. Only the first packet of each flow is registered in the database, while subsequent packets are processed using the cached hash value. This segmentation reduces the number of database registrations while maintaining complete analysis capability for all packets in the flow.
Solution Approach 2:
The patent performs preliminary action by calculating and caching the hash value of inner and outer headers when the first packet is registered. This pre-computed hash value is then reused for subsequent packets in the same flow, eliminating the need to recalculate and re-register the same associations, thereby improving processing speed.
2Measurement precision
If all database associations are accessed to retrieve outer header information, then accurate packet retrieval is achieved, but access load increases
Solution Approach 1:
The patent introduces a cache as an intermediary between the database and the packet processing system. The cache stores hash values of inner and outer headers for subsequent packets, allowing the system to retrieve header information without accessing the database for every packet. This intermediary structure reduces database access load while maintaining retrieval accuracy.
3Reliability
If inner and outer headers are associated for all packets, then complete flow tracking is achieved, but processing time increases
Solution Approach 1:
The patent segments packet processing into two distinct phases: registration phase for the first packet of each flow, and retrieval phase for subsequent packets. This segmentation allows the system to maintain complete flow tracking information while reducing processing time by avoiding redundant registration operations for subsequent packets in the same flow.
Solution Approach 2:
The patent creates a copy of the hash value association in the cache for subsequent packets. Instead of performing full database registration operations for each packet, the system copies the pre-computed hash value from the cache, significantly reducing processing time while maintaining complete flow tracking capability.
Data Source
AI summary
A registration device (10) includes an extracting unit (131) that extracts inner header information and outer header information of an encapsulated packet, a filter unit (132) that calculates a hash value of the inner header information and the outer header information as an address of a hash table in which arrival information indicating whether a first packet of a series of flow has arrived is registered for each address and causes, based on the hash table, inner header information and outer header information of the first packet of the series of flow to pass, and a registering unit (133) that registers the inner header information and the outer header information of the first packet, which the filter unit (132) has caused to pass, in a database in association with each other.