AFA Server Data Access Management via Intermediary Mirroring
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The management of accessing Flash memory in all flash array (AFA) servers is complex and often results in degraded performance due to differences between SSDs and HDDs, leading to improper control methods that can invalidate data storage solutions.
Innovation Solution
A method for data access management in AFA servers involving a dual-node architecture with upper and intermediate program modules that mirror data between nodes and send acknowledgments without checking non-volatile memory protection, enhancing performance and data protection without introducing side effects.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional Flash memory access control mechanisms are implemented in each SSD controller, then data access management conforms to specifications, but device complexity and performance degradation occur
Solution Approach 1:
The patent introduces an intermediary software layer (file system or block device driver) that mediates between the host and SSD controllers. This intermediary handles Flash memory access control policies centrally, eliminating the need for complex control mechanisms in each SSD controller. The intermediary translates host I/O operations into Flash-aware commands, managing wear leveling, garbage collection, and access patterns without requiring changes to individual SSD hardware controllers.
Solution Approach 2:
The patent extracts the Flash memory access control functionality from the SSD controllers and relocates it to the host system's software layer. By taking out the control logic from the storage devices themselves and placing it in the host's file system or driver, the system achieves compliant Flash management without burdening the SSD controllers with complex control mechanisms.
2Productivity
If SSDs are used to replace HDDs in AFA servers, then data storage performance improves, but control methods become improper or invalid due to characteristic differences
Solution Approach 1:
The patent changes the operational parameters and access patterns when interacting with SSDs compared to traditional HDDs. The intermediary software layer implements Flash-specific parameters such as page-sized I/O operations, sequential write patterns, and wear-leveling strategies. By adapting the control parameters to match Flash memory characteristics rather than using HDD-optimized methods, the system achieves both high performance and proper SSD management.
Solution Approach 2:
The patent introduces dynamic control mechanisms that adapt to SSD characteristics in real-time. The intermediary software dynamically adjusts I/O patterns, cache management strategies, and data placement based on Flash memory state, wear levels, and performance metrics. This dynamic adaptation allows the system to maintain optimal performance while respecting Flash constraints, making the control method versatile across different SSD configurations.
3Productivity
If data is mirrored to volatile memory on a second node before flushing to non-volatile memory, then acknowledgment can be sent immediately improving performance, but data protection reliability may be compromised
Solution Approach 1:
The patent performs preliminary mirroring of data to the second node's volatile memory before the write operation is considered complete. By pre-positioning data copies in fast volatile memory on a remote node, the system can immediately acknowledge the write request to the host while maintaining data protection through the mirrored copy. This preliminary action separates the acknowledgment timing from the actual persistence timing.
Solution Approach 2:
The patent creates copies of data in volatile memory on a second node as a protective measure. Instead of relying on a single point of failure, the system maintains redundant copies in fast memory that can serve as immediate recovery points. This copying strategy provides data protection without requiring synchronous flushing to non-volatile media, enabling faster acknowledgments while maintaining reliability through redundancy.
Data Source
AI summary
A method for performing data access management of an all flash array (AFA) server and the AFA server operating according to the method are provided. The method includes: utilizing an upper layer program module running on a first node to detect whether any request from a client device of a user is received; in response to a write request, utilizing an intermediate layer program module to mirror data corresponding to the write request to a second node; and before the intermediate layer program module flushing the data to a lower layer program module, in response to the data being mirrored from a first volatile memory of the first node to a second volatile memory of the second node, utilizing the intermediate layer program module to send an acknowledgement to the client device without checking whether the data has been protected in any non-volatile memory of any of the multiple nodes.