Coherency Controller Secondary Path for Interconnect Redundancy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In multiple-node systems, particularly Non-Uniform Memory Access (NUMA) systems, transaction managers may fail to access remote resources due to faults in the interconnect paths, leading to operational issues when primary communication paths are compromised.
Innovation Solution
Implementing a secondary path for coherency controllers to ensure transactions are rerouted through an alternate interconnection network in case of primary path failures, providing redundancy and maintaining system functionality.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a primary path is used for transaction processing, then system performance is optimized, but system reliability deteriorates when the path fails
Solution Approach 1:
The system pre-establishes a secondary path as a backup before the primary path fails. The coherency controller is configured with both a primary path and a secondary path to the interconnection network, allowing immediate failover without requiring path discovery or configuration changes when the primary path fails.
Solution Approach 2:
The coherency controller acts as an intermediary that manages multiple paths to the interconnection network. It monitors the health of the primary path and can switch to the secondary path through the I/O controller, providing a mediator layer that abstracts the complexity of path management from the rest of the system.
2Reliability
If additional communication links are added for redundancy, then system reliability improves, but system cost and complexity increase
Solution Approach 1:
The secondary path shares existing communication infrastructure components with the primary path, such as using the same interconnection network and I/O controller. This multi-functional use of existing resources provides redundancy without requiring entirely separate communication channels, reducing the overall complexity and cost.
Solution Approach 2:
The system creates a logical copy of the primary path functionality through the secondary path, which uses the same interconnection network resources. Rather than building a completely redundant physical infrastructure, the system copies the path logic and control mechanisms while sharing the underlying communication hardware.
Data Source
AI summary
Employing a coherency controller having a primary path and at least one secondary path to at least one interconnection network is disclosed. A method of an embodiment of the invention is performed by the coherency controller of a node. The coherency controller determines whether transactions are being properly sent to other nodes of a plurality of nodes of which the node is a part via a primary path. In response to determining that the transactions are not being properly sent to the at least one interconnection network via the primary path, the coherency controller instead sends the transactions to the other nodes via a secondary path.


