Hybrid ECC Storage for High Raw-Error-Rate Media
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Emerging storage media, such as 3D Xpoint and ReRAM, suffer from high raw-error-rates (RER) that limit their application in server systems, leading to performance issues like increased latency and write difficulty due to the inefficiencies of conventional strong error-correction coding schemes.
Innovation Solution
A hybrid storage system is implemented, using both high-RER storage media protected by weak error-correction codes and low-RER storage media protected by strong error-correction codes, with user data stored in both, and an incremental encoding mechanism to ensure real-time encoding and reduce latency by accessing low-RER media only when high-RER media errors occur.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional strong error-correction coding schemes are applied to high-RER storage media, then data reliability is improved, but write performance and throughput deteriorate significantly
Solution Approach 1:
The patent segments the error correction task by dividing data into multiple chunks and applying different ECC schemes to different portions. Specifically, it uses a first ECC scheme for first data chunks and a second, stronger ECC scheme for second data chunks, allowing the system to balance reliability and performance by applying strong error correction only where necessary rather than uniformly to all data
Solution Approach 2:
The patent changes the ECC parameters dynamically based on data characteristics. It adjusts the ECC strength parameter by selecting between a first ECC scheme and a second ECC scheme with different correction capabilities, matching the error correction strength to the actual error rate requirements of different data portions, thereby avoiding the performance penalty of uniformly applying strong ECC
2Reliability
If strong error-correction codes are used on high-RER storage media, then error correction capability is improved, but access latency increases
Solution Approach 1:
The patent segments data access operations by identifying which data chunks require strong error correction and which can use weaker schemes. This segmentation allows the system to minimize the time spent on heavy error correction operations by applying strong ECC only to specific second data chunks rather than all data, thereby reducing overall access latency while maintaining reliability
3Productivity
If high-RER storage media are used, then storage capacity and speed are improved, but data reliability deteriorates
Solution Approach 1:
The patent applies local quality by using different error correction schemes for different data chunks based on their specific requirements. Instead of applying a uniform ECC scheme across all data, it tailors the error correction strength to local conditions - using a first ECC scheme for some chunks and a second, stronger ECC scheme for others, thereby optimizing the balance between storage speed and data reliability for each portion of data
Data Source
AI summary
One embodiment provides a system and method for storing data. During operation, the system receives a to-be-written data chunk, sends the to-be-written data chunk to a first and second storage devices. The system performs first and second error-correction-code (ECC) encoding operations on the to-be-written data chunk prior to writing the to-be-written data chunk to the first and second storage media associated with the first and second storage devices, respectively. The first storage medium has a first access granularity and a first raw-error-rate (RER). The second storage medium has a second access granularity and a second RER. The first access granularity is smaller than the second access granularity, the first RER is greater than the second RER, and the second ECC encoding operation has a stronger error-correction capability than the first ECC encoding operation.