Dynamic Deduplication Switching for Storage Response Time
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing storage systems face challenges in balancing response time and processing load when employing inline and postprocess deduplication methods, particularly in scenarios where multiple write requests are sequentially transmitted from the same source, as postprocess deduplication's asynchronous and deferred processing can impact performance.
Innovation Solution
An information processing apparatus that identifies write requests from the same source within a predetermined time period and determines whether to apply a first-method deduplication process or a second-method deduplication process based on specific criteria, including the minimum interval of write requests and response times, to optimize the use of both methods and minimize processing load.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of time
If postprocess deduplication is employed, then response time to write requests is shortened, but processing load impacts performance of response processing
Solution Approach 1:
The patent dynamically selects between inline and postprocess deduplication methods based on write request patterns. When sequential write requests are detected from the same source, the system switches to postprocess deduplication to minimize response time, while for other scenarios it uses inline deduplication to maintain processing performance. This dynamic adaptation resolves the contradiction by adjusting the deduplication approach according to actual workload characteristics.
Solution Approach 2:
The system changes the operational parameters of the deduplication process by switching between two distinct modes (inline and postprocess). Each mode has different characteristics regarding response time and processing load. By changing the mode parameter based on detected write patterns, the system optimizes the balance between response time and processing performance for different operational scenarios.
2Productivity
If inline deduplication is employed, then processing load is reduced, but response time increases
Solution Approach 1:
The system dynamically adjusts the deduplication method based on the characteristics of write requests. When non-sequential write patterns are detected, inline deduplication is employed to maintain processing efficiency. When sequential patterns are identified, the system transitions to postprocess deduplication. This dynamic selection resolves the contradiction by matching the deduplication method to the actual write workload characteristics.
3Loss of time
If postprocess deduplication is used for all sequential write requests, then overall response time is shortened, but processing load increases
Solution Approach 1:
The patent applies different deduplication methods to different subsets of write requests based on their local characteristics. Instead of uniformly applying postprocess deduplication to all sequential requests, the system identifies specific patterns and applies postprocess only when beneficial, while using inline for other cases. This localized approach optimizes response time for relevant scenarios while avoiding unnecessary processing load increases.
Solution Approach 2:
The system applies postprocess deduplication partially - only to write requests that match specific sequential patterns from the same source - rather than excessively applying it to all requests. This partial application achieves response time optimization for the critical subset of requests without incurring the full processing load of universal postprocess deduplication.
Data Source
AI summary
An information processing apparatus includes an identifying unit and a determining unit. The identifying unit identifies write requests received from the same source at intervals each within a predetermined time period. The determining unit determines to perform a write operation employing first-method deduplication for a preset number of write requests selected amongst the write requests in reverse chronological order according to reception times thereof, and perform the write operation employing a second-method deduplication for the remaining write requests. The first-method deduplication includes postprocess deduplication while the second-method deduplication includes inline deduplication. The preset number is set based on a minimum interval of write requests sequentially issued from the source and time periods in the first-method deduplication and the second-method deduplication from the reception of each write request to the transmission of the response.