Mirrored Cache Data Protection Against Loss
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional computing systems face significant performance bottlenecks and data loss due to differing device performance characteristics, particularly in cache systems where a single copy of data is maintained, leading to catastrophic losses if the memory subsystem fails.
Innovation Solution
Implementing a method of cache data redundancy through mirrored copies, where data items are copied from a clean space to a dirty space in multiple cache devices, and subsequently transferred back to the data source, with the option to store multiple copies to protect against data loss and component failures.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If a single copy of cached data is maintained to improve system performance, then access speed is improved, but data loss risk increases upon memory subsystem failure
Solution Approach 1:
The patent implements mirrored copies of cached data across multiple cache devices. When data is cached in a first cache device, a copy is simultaneously maintained in a second cache device. This copying mechanism allows the system to maintain fast access speeds through caching while protecting against data loss by having redundant copies available if one cache device fails.
Solution Approach 2:
The patent divides the cache system into multiple independent cache devices, each holding copies of the same data. This segmentation allows the system to distribute data redundancy across separate physical components, so that failure of one segment (cache device) does not result in total data loss, while each segment continues to provide fast access capability.
2Reliability
If mirrored copies of cached data are maintained to protect against data loss, then data reliability is improved, but system complexity increases
Solution Approach 1:
The patent combines the data protection function with the existing cache structure by implementing mirrored copies within the same cache subsystem. Rather than adding a completely separate redundancy system, the patent merges protection into the cache architecture itself, where each cache device naturally maintains copies of data, simplifying the overall system design while achieving reliability.
Solution Approach 2:
The cache devices in the patent serve multiple functions: they provide fast data access as traditional caches do, and simultaneously provide data protection through mirrored copies. This multi-functionality reduces the need for separate protection systems, thereby limiting the increase in system complexity while achieving reliable data protection.
3Reliability
If cached data is continuously replicated across multiple cache devices, then data redundancy is improved, but memory bandwidth consumption increases
Solution Approach 1:
The patent establishes mirrored copies of cached data in advance, before any data loss event occurs. By proactively maintaining redundant copies in multiple cache devices as data is cached, the system ensures redundancy is already in place without needing to perform costly replication operations after failures or during high-demand periods, thereby reducing bandwidth consumption during critical operations.
Data Source
AI summary
Methods of protecting cache data are provided. For example, various methods are described that assist in handling dirty write data cached in memory by duplication into other locations to protect against data loss. One method includes caching a data item from a data source in a first cache device. The data item cached in the first cache device is designated with a first designation. In response to the data item being modified by a data consumer, the designation of the data item in the first cache device is re-assigned from the first designation to a second designation, and the data item with the second designation is copied to a second cache device.