Tenant-Aware Storage Sharing Engine for SSD Performance Optimization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Data centers face challenges in predicting and managing the performance characteristics of Solid State Drives (SSDs) due to their heterogeneous and time-varying latency and bandwidth, which complicates the selection and optimization of storage devices for applications with varying requirements such as redundancy, parallelism, and security.
Innovation Solution
The Tenant-Aware Storage-Sharing Engine (TASTE) is introduced, which takes into account storage resource information and application requirements to dynamically select and configure storage devices, forming virtual storage devices with optimized FTL mappings and page sizes, and providing features like replication, parallelism, and dynamic pool control to meet specific application needs.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If conventional storage management (volume management or RAID-like approaches) is used, then storage devices can be managed, but the performance characteristics of SSDs and application requirements cannot be effectively addressed
Solution Approach 1:
The patent implements dynamic storage management where the system continuously monitors SSD performance characteristics (latency, bandwidth, wear level) and adapts the storage configuration in real-time. The storage manager dynamically adjusts mapping strategies, redundancy levels, and device allocation based on current application requirements and observed SSD behavior, transforming static RAID/volume management into a dynamic, adaptive system.
Solution Approach 2:
The system changes key storage parameters (FTL mapping strategies, page sizes, cluster sizes, redundancy factors) based on application requirements and SSD performance characteristics. Different applications receive customized parameter configurations optimized for their specific needs, while the system continuously adjusts these parameters as SSD characteristics evolve over time.
2Productivity
If SSD performance characteristics are monitored and optimized, then storage efficiency improves, but the system complexity increases due to heterogeneous and time-varying characteristics
Solution Approach 1:
The patent implements continuous feedback loops where the storage manager monitors SSD performance metrics (latency, bandwidth, wear level, temperature) and uses this feedback to adjust storage operations. The system observes how SSDs actually perform under workload conditions and dynamically modifies mapping strategies, device allocation, and redundancy configurations based on observed behavior, creating a self-optimizing system.
Solution Approach 2:
The storage management system automatically monitors, analyzes, and optimizes SSD performance without requiring manual intervention. The system self-adjusts FTL parameters, dynamically allocates storage devices to applications, and manages wear leveling based on observed performance characteristics, enabling the storage subsystem to service itself and optimize performance autonomously.
3Speed
If storage devices are selected and configured dynamically, then bandwidth and latency are optimized, but the management overhead increases
Solution Approach 1:
The system performs preliminary characterization of SSD devices during initialization and idle periods, building performance profiles that predict future behavior. Application requirements are pre-analyzed and stored in the system, enabling rapid matching of applications to suitable storage devices without extensive real-time analysis. This preliminary preparation reduces the complexity of dynamic decision-making during active operations.
Data Source
AI summary
A Tenant-Aware Storage-Sharing Engine (TASTE) (225) is described. The TASTE (225) may include storage (510) for information about a set of available storage devices (705, 710) at a data center (105). A reception logic 505) may receive storage device requirements (415) from a tenant (145, 230). Selection logic (515) may then select a subset of the available storage devices (705, 710) that satisfies the storage device requirements 415) of the tenant (145, 230).