Solid State Storage Failure Prediction via ECC and Environmental Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing solid-state storage subsystems struggle to predict failures caused by conditions other than wear, as usage statistics for program/erase cycles are insufficient and vary due to manufacturing irregularities or unusual operating conditions, making it difficult to reliably estimate the timing of wear-related failures.
Innovation Solution
The storage subsystem maintains error correction code statistics and historical data on operating conditions such as temperature, altitude, humidity, and input voltage, allowing analysis to assess failure risks and trigger alerts or operational changes, while also tracking program/erase cycles.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If usage statistics regarding program/erase cycles are used to predict failures, then wear-related failure prediction is improved, but prediction accuracy for other failure conditions deteriorates
Solution Approach 1:
The monitoring system is designed to track multiple types of data simultaneously - program/erase cycle counts for wear-related failures, ECC error statistics for data integrity issues, and environmental condition data for operational stress assessment. This multi-functional approach allows a single system to predict various failure modes without requiring separate monitoring mechanisms for each failure type.
Solution Approach 2:
The failure prediction is segmented into different failure modes, each monitored by specific data types. Wear-related failures are monitored through program/erase cycle statistics, while data integrity failures are monitored through ECC error rates, and environmental failures through temperature and voltage monitoring. This segmentation allows precise prediction for each failure type without compromising overall system accuracy.
2Reliability
If comprehensive monitoring data is collected to improve failure prediction, then reliability assessment is improved, but device complexity increases
Solution Approach 1:
The storage subsystem autonomously collects and processes its own monitoring data without requiring external intervention. The controller automatically tracks program/erase cycles, monitors ECC errors, and records environmental conditions, then processes this data to generate failure risk assessments. This self-service capability improves reliability assessment while minimizing the complexity burden on the host system.
Solution Approach 2:
Multiple monitoring functions are merged into a single integrated system within the storage subsystem controller. The same controller that manages data storage operations also collects program/erase statistics, monitors ECC errors, and tracks environmental conditions, then combines all this data to produce a unified failure risk assessment. This merging reduces overall system complexity compared to having separate monitoring systems for each function.
Data Source
AI summary
A storage subsystem is disclosed that maintains (a) statistics regarding errors detected via an ECC (error correction code) module of the storage subsystem; and/or (b) historical data regarding operating conditions experienced by the storage subsystem, such as temperature, altitude, humidity, shock, and/or input voltage level. The storage subsystem, and/or a host system to which the storage subsystem attaches, may analyze the stored data to assess a risk of a failure event such as an uncorrectable data error. The results of this analysis may be displayed via a user interface of the host system, and/or may be used to automatically take a precautionary action such as transmitting an alert message or changing a mode of operation of the storage subsystem.