TCP/IP Offload Engine Failover Mechanism
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing TCP/IP offload solutions lack reliable fail-over mechanisms for maintaining continuous TCP connections when a primary TOE fails, as they assume the original system remains accessible during migration, which is not valid in case of a failure.
Innovation Solution
Implement a method where the alternate TOE retrieves periodically saved data packet communication protocol parameters from the host, recreates socket and TCP connections, and sets up a failed-over connection, ensuring seamless transition and data integrity by using a circular buffer to handle unacknowledged packets.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If TCP/IP stack functionality is offloaded from Host to TOE, then CPU bandwidth on Host is freed, but reliability of TCP connections is reduced due to lack of fail-over mechanism
Solution Approach 1:
The patent implements preliminary checkpointing of TCP connection parameters to the Host before TOE failure occurs. The primary TOE periodically saves critical TCP state information (socket parameters, connection state, sequence numbers) to the Host system memory, enabling the alternate TOE to restore connections without dropping them when failure occurs.
Solution Approach 2:
The patent creates a copy of TCP connection state information from the primary TOE to the Host system. The checkpointed parameters include complete TCP socket state that can be replicated by the alternate TOE, allowing it to assume the primary TOE's connections seamlessly without requiring application-level migration.
2Device complexity
If alternate TOE is kept passive with no data traffic, then system complexity is reduced, but fail-over capability is insufficient for reliable connection continuity
Solution Approach 1:
The patent enables the alternate TOE to autonomously restore TCP connections by retrieving checkpointed parameters from the Host and recreating socket states without requiring active participation during normal operation. The passive TOE becomes active automatically upon detecting primary TOE failure, taking over connections independently.
Solution Approach 2:
The patent implements a feedback mechanism where the primary TOE periodically checkpoints connection parameters to the Host, creating a backup state that feeds into the fail-over capability. This continuous parameter saving provides the alternate TOE with the necessary information to restore connections when switching occurs.
3Reliability
If TCP connection migration is implemented assuming original system remains accessible, then connection continuity is maintained, but the solution becomes complex and invalid when original system fails
Solution Approach 1:
The patent extracts the essential TCP connection state information from the TOE and stores it in the Host system memory. This separation allows the connection state to be preserved independently of any specific TOE, enabling the alternate TOE to restore connections without needing the original TOE to remain accessible or participate in the migration process.
Data Source
AI summary
A computing system, comprising a host computer system, a primary data packet communication protocol (such as TCP/IP) offload engine configured to act as an offload engine for the host computer system, and an alternate data packet communication protocol (such as TCP/IP) offload engine configured to act as an offload engine for the host computer system if the primary offload engine fails. The host computer system, the primary offload engine and the alternate offload engine are in data communication, the primary offload engine is configured to save predefined data packet communication protocol parameters to the host computer system during normal operation, the alternate offload engine is configured to retrieve the parameters from the host computer system if the primary offload engine fails.