Application Data Reordering for Faster Interactive Streaming Startup
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing application streaming methods face challenges such as long download times, high latency, and unsatisfactory user experiences due to large application sizes, especially in interactive applications like games, which hinder the 'try before you buy' market and result in unacceptable delays before applications can be interactively used.
Innovation Solution
A method for generating re-ordered application data by analyzing data chunk access patterns from multiple application sessions to optimize data streaming, utilizing two independent download mechanisms and controlling file I/O to ensure seamless application execution with minimal initial data, allowing applications to start quickly and maintain performance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the application is downloaded completely from a remote server before execution, then the application can run without data availability issues, but the download time becomes excessively long and unacceptable to users
Solution Approach 1:
The patent reorders application data chunks based on their expected access patterns during execution, so that the most frequently and early-accessed data is downloaded first. This preliminary organization of data allows the application to start executing with a smaller initial download while maintaining reliability by pre-positioning critical data.
Solution Approach 2:
The application data is divided into discrete chunks that can be downloaded independently. By segmenting the data and ordering these segments according to access patterns, the system enables partial downloads to be sufficient for starting the application, reducing overall download time while maintaining functionality.
2Loss of time
If data chunks are downloaded until an executable threshold is reached and the application is started while remaining data chunks are downloaded in the background, then the start time is reduced, but the pre-load data required is still large and difficult to determine
Solution Approach 1:
The patent uses trace statistics information collected from multiple application sessions to determine the expected access patterns of data chunks. This feedback mechanism allows the system to automatically determine the optimal pre-load data requirement based on actual usage patterns, reducing the complexity of manual determination and enabling more accurate threshold setting.
Solution Approach 2:
By analyzing trace data from previous sessions, the system preliminarily determines the access patterns and orders data chunks accordingly before the actual application execution. This preliminary analysis enables more accurate determination of the executable threshold with smaller pre-load requirements.
3Productivity
If the application data is reordered based on expected access patterns, then the download efficiency is improved and start time is reduced, but additional processing is required to analyze trace statistics and determine occurrence positions
Solution Approach 1:
The trace statistics information and data chunk occurrence positions are determined in advance during offline analysis of application sessions. This preliminary processing separates the complex analysis work from the actual application startup process, allowing the runtime system to simply follow the pre-determined optimal data order without performing complex real-time analysis.
4Measurement precision
If trace statistics information is collected from multiple application sessions to determine data chunk occurrence positions, then the data ordering accuracy is improved, but the data collection and processing time increases
Solution Approach 1:
The trace data collection and processing is segmented into offline batch processing of multiple sessions, rather than requiring real-time analysis. This allows accurate statistics to be gathered from multiple sessions without impacting the startup time of individual application instances, as the analysis is performed separately in advance.
Data Source
AI summary
The disclosure inter alia pertains to methods and devices for generating re-ordered application data from original application data of an application, for recording a trace of an application session and for running an application.