Object Addressable Storage Bypass for Block I/O Latency
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing systems that combine block I/O storage systems with object addressable storage interfaces face inefficiencies due to content units being transferred twice over networks, increasing latency and network traffic, especially when dealing with large content units.
Innovation Solution
A method where a first server issues object identifiers to a second server, which converts them into block I/O addresses for a block I/O storage system, allowing direct access and transfer of content units from the storage system to the host computer, bypassing the object addressable storage server.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If content units are transferred through the object addressable storage server, then the system maintains location independence and object identifier abstraction, but network latency and traffic increase due to double transfer
Solution Approach 1:
The patent extracts the content unit directly from the block I/O storage system and delivers it to the host computer, bypassing the object addressable storage server for the actual data transfer. Only the object identifier to address mapping goes through the second server, while the content itself takes a direct path, eliminating unnecessary network hops and reducing latency
Solution Approach 2:
The system segments the access process into two independent parts: (1) object identifier to address conversion through the second server, and (2) direct content transfer from storage system to host computer. This segmentation allows the metadata operation to go through the OAS server while the bulk data transfer bypasses it, optimizing both abstraction maintenance and transfer efficiency
2Adaptability or versatility
If content units are transferred through the object addressable storage server, then the system maintains object addressable interface, but network traffic increases especially for large content units
Solution Approach 1:
The patent extracts the content unit transfer path from the object addressable storage server and creates a direct connection between the block I/O storage system and the host computer. This extraction eliminates the server as an intermediate node for data transmission, significantly reducing network traffic while preserving the object addressable interface through the identifier-to-address conversion mechanism
3Speed
If block I/O storage system is used directly, then access speed is fast, but location independence and object identifier abstraction are lost
Solution Approach 1:
The patent introduces a second server as an intermediary that performs object identifier to block I/O address conversion. This mediator enables the host computer to use object identifiers for access while the actual storage operations use efficient block I/O addresses, combining the benefits of location independence with fast direct access speeds
Solution Approach 2:
The system segments the access functionality into two parts: the second server handles the abstraction layer (object identifier management and address conversion), while the block I/O storage system handles the performance-critical data access operations. This segmentation allows each component to optimize for its specific function
Data Source
AI summary
One embodiment of the invention relates to the transfer of content between a host computer that issues OAS access requests and a block I/O storage system. Specifically, a host computer may issue an access request for a content unit that identifies the content unit is an object identifier. The request may be received by a second server, which may determine the block address(es) on the block I/O storage system at which the content unit is stored. A request may then be sent to the block I/O storage system to retrieve the content stored at the requested block address(es) and the block I/O storage system may return the content.