Hybrid SSD Spinning Disk Storage API for Adaptive Streaming
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing storage solutions for HTTP adaptive streaming face challenges in scaling performance due to the mix of large and small files, leading to reduced throughput and high costs, especially when using spinning disks, and complex solutions like SSDs introduce latency and increased complexity.
Innovation Solution
The solution combines solid state disks (SSDs) and spinning disks by transparently allocating content based on size, with small objects stored on SSDs for efficient access and large objects on spinning disks, using a single storage API to manage both without requiring changes to the delivery application.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If spinning disks are used for storage, then storage capacity is achieved, but throughput performance is severely reduced due to head positioning overhead for mixed file sizes
Solution Approach 1:
The patent segments the storage system into two distinct parts: a fast storage layer (SSD) for small files and a capacity storage layer (spinning disk) for large files. This segmentation allows each layer to be optimized for its specific use case, with the SSD handling small file I/O operations that would otherwise bottleneck the spinning disk system, thereby improving overall throughput without requiring complete system redesign
Solution Approach 2:
The patent applies local quality by assigning different storage media to different file size categories. Small files (typically configuration files, manifests, and metadata) are stored on SSD with high I/O performance characteristics, while large files (video segments) are stored on spinning disks with high capacity characteristics. This localized optimization ensures that each storage medium operates in its performance sweet spot
2Speed
If solid state disks (SSD) are used, then access performance for small objects is improved, but cost increases by 3-4 times compared to spinning disks
Solution Approach 1:
The patent changes the parameter of file size classification to determine storage location. By establishing a size threshold (e.g., files smaller than a certain size go to SSD, larger files go to spinning disk), the system dynamically adjusts which storage medium is used based on file characteristics. This parameter-based differentiation allows the system to achieve high access speeds for small files on SSD while maintaining cost-effective spinning disk storage for large files
Solution Approach 2:
The patent creates a composite storage system that combines SSD and spinning disk technologies into a unified storage architecture. This composite approach leverages the strengths of both media types: the high I/O performance of SSD for small files and the high capacity and lower cost of spinning disks for large files, achieving a cost-performance balance that neither medium could achieve alone
3Productivity
If small objects are combined into big files, then disk performance is improved, but application complexity and memory consumption increase
Solution Approach 1:
Instead of combining small files into large files to improve disk performance (the conventional approach), the patent inverts the strategy by keeping small files separate and storing them on SSD. This inversion eliminates the need for complex file combination logic in the application layer, as the storage system naturally handles small files efficiently through the SSD layer, thereby improving disk performance without increasing application complexity
Data Source
Figure 1
AI summary
A method and server (10) for storing multimedia content from an origin node (16) and delivering multimedia content in HTTP adaptive streaming to an end-user device (17) requesting files of multimedia content are proposed, wherein a single storage API (13) is used for writing files of multimedia content to be stored and reading files of multimedia content to be delivered up to an adaptive streaming delivery application layer; comprising: - receiving multimedia content and converting the received content into files of a determined size, - sending a command to the API (13) for writing each file, depending on its size, either into a first filesystem (11) of solid state disks or into a second filesystem (12) of spinning disks, wherein both the first and second filesystems (11, 12) communicate with the API (13). The delivery application layer over the API (13) is unaware of both solid state and spinning disks filesystems (11, 12).