Disaggregated Tiered Cache Bandwidth Allocation for Shared Cloud Storage
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing cloud content storage systems face limitations in efficiently utilizing storage resources due to processor saturation, leading to underutilization of local storage resources despite available bandwidth and capacity.
Innovation Solution
Implementing a disaggregated tiered cache system where a storage device pool is shared among multiple hosted devices, allowing game requests to be distributed efficiently across suitable storage devices, with bandwidth management and namespace allocation to optimize storage resource utilization.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If a centralized storage system is used to serve content requests, then content delivery can be simplified, but storage resource utilization efficiency deteriorates due to processor saturation and inability to distribute load
Solution Approach 1:
The storage system is segmented into multiple independent storage devices (e.g., SSDs, HDDs) that can be individually accessed and managed. Each storage device operates as an independent unit within the pool, allowing parallel content delivery to multiple hosted devices without creating processor saturation bottlenecks.
Solution Approach 2:
The storage device pool serves multiple functions simultaneously: it stores content for multiple hosted devices, distributes content requests across different storage devices, and manages bandwidth allocation dynamically. This multi-functional approach maximizes resource utilization efficiency.
2Productivity
If local storage resources are dedicated to single hosted devices, then access speed can be optimized, but overall storage resource utilization deteriorates due to underutilization when processors are saturated
Solution Approach 1:
The system dynamically assigns storage devices to hosted devices based on real-time bandwidth availability and content requests. When a hosted device has high bandwidth availability, more storage devices can be assigned to it, optimizing both utilization and access speed under varying load conditions.
Solution Approach 2:
Multiple storage devices are merged into a single shared pool that serves all hosted devices. This combining approach allows the system to aggregate storage capacity and bandwidth resources, achieving higher overall utilization while maintaining fast access speeds through parallel operations.
3Productivity
If bandwidth is allocated to multiple hosted devices simultaneously, then system throughput improves, but storage resource allocation complexity increases due to need for bandwidth management and namespace partitioning
Solution Approach 1:
The storage namespace is segmented into separate partitions or virtual namespaces for each hosted device. This segmentation allows independent bandwidth management for each device while maintaining overall system throughput, as each hosted device can be served without interfering with others.
Solution Approach 2:
A bandwidth manager component acts as an intermediary between the storage device pool and hosted devices. This mediator handles bandwidth allocation, priority management, and namespace partitioning automatically, reducing the operational complexity of managing multi-device bandwidth distribution.
Data Source
AI summary
A content provider system includes: a repository to store a catalog of content; a storage device pool to load content from among the catalog of content from the repository into one or more storage devices of the storage device pool; a first hosted device communicably connected to the storage device pool, and to execute the content stored in the storage device pool to provide the content to a first user device; a second hosted device communicably connected to the storage device pool, and to execute the content stored in the storage device pool to provide the content to a second user device; and one or more processing circuits to identify an available storage device from among the one or more storage devices of the storage device pool for serving a requested content to a requesting device from among the first and second hosted devices.