Self-Organizing Workload Distribution in Multi-Tenant Cloud
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional approaches to managing workloads in dynamic cloud environments face challenges in efficiently distributing workload across resources, leading to inefficiencies and potential downtime due to unaccounted workload portions and server availability issues.
Innovation Solution
A system where computing devices announce and claim workload portions, with a data store maintaining key-value entries for workload distribution, and a mechanism for workload rebalancing among devices to ensure even distribution and handle unresponsive devices, allowing for self-organization and dynamic resource allocation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If load balancing is used to distribute workloads across resources, then workload distribution is improved, but the dynamic nature of demands causes difficulties in meeting varying requirements
Solution Approach 1:
The patent implements dynamic workload distribution by allowing computing devices to continuously announce their availability and claim workload portions based on real-time conditions. The system transitions from static load balancing to a dynamic mechanism where devices can join, leave, or change capacity, and workload claims are made and reassigned accordingly, enabling the system to adapt to varying demands automatically
Solution Approach 2:
The patent enables computing devices to self-manage workload distribution through autonomous announcement and claiming mechanisms. Each device independently announces its availability and capacity, and workload portions are automatically claimed by devices without requiring centralized control, allowing the system to self-adjust to dynamic conditions and resource availability
2Productivity
If computing devices independently claim workload portions, then workload distribution is improved, but unaccounted workload portions and server availability issues occur
Solution Approach 1:
The patent implements a feedback mechanism where computing devices continuously announce their availability status and claimed workload portions to the system. This announcement mechanism provides feedback about device capacity and workload assignment, enabling the system to track accounted workload portions and detect when devices are unavailable or overloaded, thereby improving reliability while maintaining efficient distribution
Data Source
AI summary
Approaches are described for managing workload, or other tasks in a data center, shared resource environment or other such electronic environment. In particular, a customer (or other end user, etc.) is able to use a computing device to submit a request to process information across at least one appropriate network to be received by a provider environment. The provider environment includes a set of resources (e.g., computing devices) operable to process the information for any of a variety of different purposes. Code can be deployed and executed at one or more of the host machines, wherein when executed, can enable the host machines to perform operations to process a workload in a distributed self-organizing manner, without a master server or other management device, to distribute work, handle situations where host machines go offline, etc.