Application Migration Between Dedicated Processing Units
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing GPU deployment in cloud environments faces challenges in dynamically migrating applications between processing units during execution, making it difficult to optimize resource utilization and quality of service, which hampers overall system performance.
Innovation Solution
A method and machine for application migration that involves obtaining descriptors associated with the runtime environment of an application, creating a mapping between these descriptors to redirect requests from one processing unit to another, allowing transparent migration without disrupting the application's execution.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If applications are statically assigned to dedicated processing units, then system stability is maintained, but resource utilization and quality of service deteriorate when load conditions change
Solution Approach 1:
The patent introduces a migration manager as an intermediary component that coordinates application migration between dedicated processing units. The migration manager obtains descriptors from both source and target processing units, establishes mappings between them, and manages the migration process, thereby reducing the complexity burden from the overall system while enabling adaptable migration.
Solution Approach 2:
The patent uses descriptor copying and mapping mechanisms where descriptors representing processing unit states are copied and mapped between source and target units. This allows the system to maintain state information during migration without requiring complex direct state transfer, simplifying the migration management process.
2Productivity
If application migration is implemented during execution, then resource utilization and quality of service improve, but system complexity and migration overhead increase
Solution Approach 1:
The patent performs preliminary actions by obtaining descriptors from processing units before migration execution and establishing mappings in advance. This preparation work is done before the actual application migration, allowing the migration to proceed more efficiently with reduced overhead during the critical migration window.
Solution Approach 2:
The patent enables continuous application execution during migration by maintaining descriptor mappings that allow the application to continue running while transitioning between processing units. The migration occurs without stopping the application, ensuring continuity of useful action and minimizing migration overhead time.
3Ease of operation
If descriptor mapping is established for migration, then application transparency is maintained, but system complexity and memory operations increase
Solution Approach 1:
The migration manager acts as an intermediary that handles the complex descriptor mapping operations, shielding the application from the underlying complexity. The application simply interacts with the migration manager, which manages the descriptor mappings between source and target processing units, maintaining transparency while managing complexity centrally.
Data Source
AI summary
Embodiments provide a method and machine for application migration. In an embodiment, a first descriptor associated with a first runtime environment of an application executed on a first dedicated processing unit is obtained from the first dedicated processing unit. In response to determining that the application is to be migrated from the first dedicated processing unit to the second dedicated processing unit during the execution, the second dedicated processing unit is caused to create a second descriptor associated with a second runtime environment for the application, executed on the second dedicated processing unit. Subsequently, the second descriptor is obtained from the second dedicated processing unit and a first mapping from the first descriptor to the second descriptor is established for re-directing, to the second dedicated processing unit, the request from the application to the first dedicated processing unit. A corresponding machine for implementing the method is further disclosed.