Asynchronous Synchronous Replication Priority Mechanism
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional data replication technologies face challenges in balancing the latency requirements of synchronous replication requests with the bandwidth needs of asynchronous replication requests, often leading to suboptimal processing where either synchronous or asynchronous requests are compromised.
Innovation Solution
A method and system that utilize an asynchronous and synchronous transmit priority mechanism, employing a machine learning system to optimize the execution order of replication requests, assigning requests to separate queues and tuning the polling frequency to balance latency and bandwidth requirements, ensuring efficient processing of both synchronous and asynchronous replication requests.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of time
If synchronous replication requests are prioritized to reduce latency, then synchronous replication performance is improved, but asynchronous replication bandwidth is reduced
Solution Approach 1:
The patent segments replication requests into two separate queues: a first queue for synchronous replication requests and a second queue for asynchronous replication requests. This segmentation allows independent processing of each queue type, enabling the system to optimize for synchronous latency by quickly draining the first queue while simultaneously optimizing for asynchronous bandwidth by efficiently processing the second queue, thereby resolving the resource contention between the two replication modes.
2Productivity
If asynchronous replication requests are prioritized to increase bandwidth, then asynchronous replication performance is improved, but synchronous replication latency increases
Solution Approach 1:
The patent implements dynamic polling frequency adjustment for the two queues. The system dynamically determines a polling frequency value that balances the competing demands: when synchronous requests are pending, the polling frequency for the first queue is increased to reduce latency; when asynchronous requests dominate, the polling frequency for the second queue is increased to maximize bandwidth. This dynamic adjustment allows the system to adapt to varying workloads and resolve the contradiction between latency and bandwidth optimization.
Data Source
AI summary
A method is used in managing an asynchronous and synchronous transmit priority mechanism. The method receives a plurality of replication requests in conjunction with a replication process, where the plurality of replication requests comprises at least one of a first replication request or a second replication request. A first storage system that comprises a plurality of storage devices is configured to participate in the replication process with a second storage system that comprises a plurality of storage devices. The method determines an execution order to optimize processing of the plurality of replication requests, where a first replication request execution order that optimizes processing of the at least one first replication request conflicts with optimizing processing of the at least one second replication request.