Module Affinity Link Selection for Synchronous Storage Replication
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Synchronous replication in information processing systems incurs significant delays and resource penalties due to increased link distance, write size, and network load, particularly in host writes, which existing technologies fail to adequately address.
Innovation Solution
Implementing 'module affinity link selection' by choosing links associated with processing modules that share the same process or processor, enabling local memory copy operations for efficient data transfer during synchronous writes, thereby reducing latency and conserving network bandwidth.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If synchronous replication is implemented to ensure data consistency between source and target storage systems, then data reliability is improved, but host write performance deteriorates due to network latency and link distance
Solution Approach 1:
The patent segments the data path by identifying specific processing modules (source processing module, transmit processing module, receive processing module, target processing module) and their associated links. By segmenting the replication path and selecting optimal segments, the system reduces the effective distance data must travel across the network, thereby improving host write performance while maintaining data consistency through the synchronous replication mechanism.
Solution Approach 2:
The patent introduces link selection as an intermediary mechanism between the source and target storage systems. By selecting optimal links based on processing module affinity, the system mediates the data transmission path to minimize network latency. This intermediary layer allows the system to maintain synchronous replication guarantees while reducing the performance penalty through intelligent path selection.
2Adaptability or versatility
If data is transmitted over long link distances between source and target storage systems, then disaster recovery capability is improved, but network bandwidth consumption increases
Solution Approach 1:
The patent applies local quality by optimizing the transmission path locally between specific processing modules rather than using a generic long-distance path. By selecting links that have affinity with the processing modules (i.e., links locally associated with the same process or processor), the system reduces network bandwidth consumption for synchronous replication while maintaining the ability to replicate data across long distances for disaster recovery.
Solution Approach 2:
The patent changes the parameter of link selection from a static, arbitrary choice to a dynamic selection based on processing module affinity. By changing how links are selected (using process identifier and processor identifier matching), the system optimizes network bandwidth utilization for disaster recovery replication without sacrificing the adaptability to replicate across different geographical distances.
3Device complexity
If traditional link selection is used for synchronous replication, then system simplicity is maintained, but write acknowledgment latency increases with link distance and network load
Solution Approach 1:
The patent performs preliminary action by pre-establishing affinity relationships between processing modules and links based on process and processor identifiers. This preliminary mapping allows the system to quickly select optimal links during synchronous replication without complex real-time calculations, thereby reducing write acknowledgment latency while adding minimal complexity through pre-computed affinity data structures.
Solution Approach 2:
The patent introduces dynamics by making link selection adaptive rather than static. The system dynamically selects links based on the specific source processing module handling each write request, using its process and processor identifiers to find the optimal transmit processing module. This dynamic selection reduces write acknowledgment latency by adapting to the actual system state while maintaining reasonable complexity through structured affinity matching.
Data Source
AI summary
An apparatus includes a processing device comprising a processor and a memory. The processing device is configured, in conjunction with synchronous replication of a logical storage volume between first and second storage systems, to receive a synchronous write request comprising at least a portion of a data page to be written to the storage volume, to determine a source processing module associated with the data page in the first storage system, to determine at least one of a process identifier and a processor identifier of the source processing module in the first storage system, and to select a particular one of multiple links between the first and second storage systems for use with the synchronous write request based at least in part on at least one of the process identifier and the processor identifier of the source processing module. The selected link is illustratively associated with a transmit processing module.