Containerized Workload Mapping Across Coherent Multi-Architecture Processors
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing coherent computing approaches struggle with dynamically allocating workloads to containers based on processor architecture conditions and fail to address data consistency issues in containerized workloads, leading to potential data corruption and lag when multiple processor architectures access shared memory simultaneously.
Innovation Solution
A computing device maps events to containers configured for different processor architectures, using remote procedure calls (RPCs) to ensure data consistency by implementing locks on shared memory and utilizing RPC middleware for inter-process communication, allowing efficient workload allocation based on processor characteristics and events.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If multiple processor architectures access shared memory simultaneously in containerized workloads, then computing versatility and resource utilization are improved, but data consistency and system reliability deteriorate due to potential data corruption and lag
Solution Approach 1:
The patent introduces an RPC (Remote Procedure Call) layer as an intermediary between containers running on different processor architectures. This RPC layer handles inter-process communication and memory access coordination, acting as a mediator that ensures data consistency while allowing diverse architectures to access shared memory. The RPC mechanism serializes access to shared memory regions, preventing simultaneous conflicting accesses that would cause data corruption.
Solution Approach 2:
The system performs preliminary actions by establishing lock mechanisms and memory barrier protocols before multiple processor architectures access shared memory. The RPC layer pre-configures access rights, memory mapping, and synchronization primitives before workload execution begins. This preliminary setup ensures that when multiple architectures access memory, the pre-established protocols maintain data consistency without requiring complex runtime negotiations.
2Productivity
If workload allocation is dynamically adjusted based on processor architecture conditions, then resource utilization and productivity are improved, but system complexity increases due to event mapping and container orchestration requirements
Solution Approach 1:
The patent implements a universal event mapping mechanism that works across different processor architectures and container types. The event mapping system uses architecture-agnostic event descriptors and container identification schemes that can represent any processor type (x86, ARM, RISC-V, etc.) and any container configuration. This universal approach allows dynamic workload allocation without requiring architecture-specific orchestration logic, reducing overall system complexity while maintaining high resource utilization.
Solution Approach 2:
The system enables self-service through automated event-to-container mapping and workload allocation. The RPC layer and event mapping mechanism automatically match incoming events to appropriate containers based on processor architecture compatibility and current system state, without requiring manual intervention or complex external orchestration. This self-service capability dynamically optimizes resource utilization while keeping the control logic distributed and manageable.
3Reliability
If locks are implemented on shared memory address ranges, then data consistency is improved, but access speed and productivity deteriorate due to potential locking overhead
Solution Approach 1:
The patent implements fine-grained locking where different address ranges in shared memory have independent lock states. Instead of locking entire memory regions, the system locks only specific address ranges that are currently being accessed. This local quality approach allows other parts of the memory space to be accessed simultaneously by different RPC calls, maintaining data consistency for locked regions while preserving access speed for unlocked regions. The granularity of locks matches the actual data access patterns of containerized workloads.
Data Source
AI summary
Aspects of the present disclosure relate to coherent containerized computing. More specifically, a method is described that includes obtaining an indication of a workload and an indication of an event. The method further includes mapping the event to a first container in a plurality of containers, where each container in the plurality of containers is configured for a different processor architecture, and where the first container is configured for a first processor architecture. The method also includes performing, by a first processing device configured with the first processor architecture, the workload by way of the first container.