Dynamic Reliability Scoring for Distributed Storage
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current distributed data storage systems lack dynamic metrics to assess reliability in real-time, relying on static measures like PDL and MTTDL, which do not account for changing device failures and reconstructions, and require high redundancy to ensure data integrity.
Innovation Solution
The introduction of a dynamic backward-looking metric called Normalcy Deviation Score (NDS) and a scheduling policy called Minimum Intersection (MinI) to dynamically quantify and improve system reliability by prioritizing reconstructions based on redundancy levels and optimizing reconstruction schedules.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If additional device redundancy is used to improve reliability, then system reliability is improved, but system cost and device complexity increase
Solution Approach 1:
The patent applies dynamics by transitioning from static reliability metrics (PDL, MTTDL) to a dynamic reliability metric (NDS) that continuously updates based on system state changes. The NDS reflects current system reliability in real-time, enabling dynamic failure management decisions without requiring additional hardware redundancy.
Solution Approach 2:
The patent changes the parameter used for reliability measurement from static probability-based metrics to a dynamic score-based metric (NDS) that incorporates time-sensitive system state information. This parameter change enables reliability assessment without additional redundancy hardware.
2Reliability
If additional device redundancy is used to improve reliability, then system reliability is improved, but expense increases
Solution Approach 1:
The system uses its existing redundancy mechanisms and operational state information to self-assess reliability through the NDS metric. The failure management policy leverages current system conditions (failures, reconstructions, replacements) to make intelligent decisions without requiring additional redundancy devices.
Solution Approach 2:
The patent implements feedback by continuously monitoring system state changes (device failures, data reconstructions, replacements) and using this information to update the NDS reliability metric. This feedback loop enables dynamic adjustment of failure management policies based on actual system conditions.
3Device complexity
If static reliability metrics (PDL, MTTDL) are used to assess system reliability, then reliability assessment is simplified, but time sensitivity and real-time accuracy are lost
Solution Approach 1:
The patent transforms static reliability metrics into a dynamic metric (NDS) that automatically incorporates time-sensitive system state information. The NDS updates in real-time based on failures, reconstructions, and replacements, providing both simplicity and time-awareness in reliability assessment.
Solution Approach 2:
The patent performs preliminary actions by pre-defining the NDS calculation framework and continuously preparing system state information for reliability assessment. This allows real-time reliability evaluation without complex computations at the moment of assessment.
Data Source
AI summary
Data is stored in a distributed data storage system comprising a plurality of disks. When a disk fails, system reliability is restored by executing a set of reconstructions according to a schedule. System reliability is characterized by a dynamic Normalcy Deviation Score. The schedule for executing the set of reconstructions is determined by a minimum intersection policy. A set of reconstructions is received and divided into a set of queues rank-ordered by redundancy level ranging from a lowest redundancy level to a highest redundancy level. For reconstructions in each queue, an intersection matrix is calculated. Diskscores for each disk are calculated. The schedule for the set of reconstructions is based at least in part on the intersection matrices, the Normal Deviation Scores, and the diskscores.