Partitioned Operating System for Predictable Distributed Computing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing virtualized environments face challenges in designing distributed computer systems that ensure predictable use of shared computing and networking resources, particularly for time-critical and safety-critical applications, as they are affected by varying workloads and network loads, leading to performance deterioration and unpredictable quality of service.
Innovation Solution
A new computer architecture that combines a communication network allowing resource partitioning and a partitioned operating system for virtualization, utilizing TTEthernet for predictable communication and integrating time-driven and dynamically scheduled partitions to provide guaranteed quality of service, while maximizing resource utilization.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If virtualization is used to share computing resources among different applications, then resource utilization is improved, but quality of service becomes unpredictable under varying workloads
Solution Approach 1:
The patent segments computing resources into distinct partitions (time partitions, space partitions, resource partitions) that are isolated from each other. This segmentation ensures that applications in different partitions cannot interfere with each other, maintaining predictable quality of service while still allowing multiple applications to share the same physical hardware infrastructure, thus resolving the contradiction between resource utilization and service predictability.
Solution Approach 2:
The patent implements preliminary resource allocation and reservation mechanisms where computing and networking resources are pre-assigned to specific partitions before execution begins. This preliminary action ensures that time-critical applications receive guaranteed resource allocations regardless of overall system workload, maintaining QoS predictability while enabling efficient resource sharing across partitions.
2Productivity
If communication resources are shared among multiple applications, then network resource utilization is improved, but end-to-end latency becomes variable and unpredictable
Solution Approach 1:
The patent segments communication resources into dedicated virtual networks and time slots for different partitions. By separating communication paths and assigning specific time windows to each partition, the system ensures predictable end-to-end latency for time-critical applications while still allowing efficient use of network infrastructure across all partitions, thus resolving the contradiction between network utilization and latency predictability.
Solution Approach 2:
The patent implements periodic time-triggered communication schedules where data transmission occurs at predetermined time intervals rather than continuously. This periodic action ensures that time-critical communications receive guaranteed time slots with predictable latency, while non-critical traffic can utilize remaining network capacity, achieving both latency predictability and network resource utilization.
3Reliability
If time-critical applications are executed with guaranteed QoS, then service reliability is improved, but hardware resource utilization decreases
Solution Approach 1:
The patent divides hardware resources into separate partitions for time-critical and non-time-critical applications. By isolating time-critical applications in dedicated partitions with guaranteed resource allocations, the system ensures service reliability for these applications. Simultaneously, non-time-critical applications can utilize remaining hardware resources, thereby maintaining high overall hardware utilization while preserving QoS guarantees for critical services.
Solution Approach 2:
The patent implements dynamic resource allocation where the system can adaptively assign hardware resources to different partitions based on current workload conditions. This dynamic mechanism allows the system to guarantee resources to time-critical applications when needed while efficiently utilizing available hardware capacity for non-critical applications during lower-demand periods, resolving the contradiction between service guarantees and hardware utilization.
4Reliability
If partitions are executed independently of system workload, then quality of service is improved, but system complexity increases
Solution Approach 1:
The patent segments the operating system into independent partitions that manage their own execution and resource allocation. Each partition operates independently with its own resource guarantees, which simplifies the overall system architecture by allowing modular design and independent optimization of each partition rather than requiring complex centralized coordination, thus resolving the contradiction between QoS reliability and system complexity.
Data Source
AI summary
The invention relates to a method for providing guaranteed quality of service in distributed computing platforms for execution of distributed applications, by combining: i) a partitioned operating system (POS) or a hypervisor (101) executed on at least two computers (300,301); where said POS/hypervisor allocates the CPU, memory and I/O hardware resources to computer partitions in said computers; where said computers are capable of executing different computer partitions (104, 105, 106) with different application tasks within one said computer partition or executing different operating systems (in case of hypervisor) within one said computer partition; wherein the said POS/hypervisor ensures that the application tasks executed in different partitions within the same said computer get their allocated hardware resources; wherein said POS/hypervisor performs a time-driven scheduling for a subset of computer partitions based on a given configuration (136), ii) means, for example network communication technology, for implementing partitioning of Ethernet communication resources and creating virtualized networks out of one physical network.