Dynamic Resource Allocation in Geo-Distributed Cloud Infrastructures
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing cloud-based datacentre systems face inefficiencies and increased latency when client requirements change, as traditional approaches require continuous reconfiguration of virtual datacentre components across multiple physical datacentres, leading to communication inefficiencies and complex reconfiguration processes.
Innovation Solution
A controller for geo-distributed cloud infrastructures that dynamically reallocates client systems to physical datacentres based on existing capabilities, minimizing reconfiguration overhead and maintaining existing dependencies between virtual datacentre components, thereby optimizing service delivery and reducing latency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If traditional approaches continuously reconfigure virtual datacentre components across multiple physical datacentres to meet changing client requirements, then service adaptability is improved, but latency increases and communication efficiency deteriorates
Solution Approach 1:
The system segments the cloud infrastructure into Virtual Service Cells (VSCs), each serving as an autonomous unit that can be independently allocated to client systems. This segmentation allows flexible service adaptation within each cell while avoiding the need for continuous cross-datacentre reconfiguration, thereby reducing latency while maintaining adaptability.
Solution Approach 2:
The system performs preliminary allocation of Virtual Service Cells to client systems based on predicted or initial requirements. By pre-establishing these allocations and only performing reallocation when absolutely necessary, the system avoids continuous reconfiguration overhead and the associated latency, while still maintaining service adaptability through the ability to reallocate when needed.
2Adaptability or versatility
If virtual datacentre components are distributed across multiple physical datacentres to meet changing requirements, then service versatility is improved, but device complexity increases
Solution Approach 1:
By dividing the system into discrete Virtual Service Cell units, the patent simplifies the management of service versatility. Each VSC can be independently allocated or reallocated as a complete unit, avoiding the complexity of managing individual components across multiple datacentres. This segmentation maintains service versatility while reducing reconfiguration complexity.
Solution Approach 2:
The controller acts as an intermediary that manages the allocation and reallocation of Virtual Service Cells. It handles the complexity of service versatility requirements by abstracting the underlying datacentre infrastructure, presenting a simplified interface for resource management, and automatically handling the complex reconfiguration processes only when necessary.
3Adaptability or versatility
If continuous reconfiguration is performed to meet changing demand profiles, then service adaptability is improved, but reconfiguration overhead and operational complexity increase
Solution Approach 1:
Instead of continuous reconfiguration, the system performs periodic or event-driven reallocation of Virtual Service Cells based on changing demand profiles. This periodic action reduces reconfiguration overhead by only performing allocations when necessary, while still maintaining service adaptability through timely responses to demand changes.
Solution Approach 2:
The controller automatically manages the reallocation of Virtual Service Cells in response to changing demand profiles without requiring manual intervention. This self-service capability reduces operational complexity by automating the entire reallocation process, from detecting demand changes to executing the reallocation, thereby improving reconfiguration efficiency while maintaining service adaptability.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
As data processing requirements of individual client systems A to K change over time, they are allocated service from physical datacentres X, Y, Z according to the existing capabilities of the data centres, thus transferring client systems between cells controlled by individual data centres. This avoids the complex process of adapting the individual physical datacentres' capabilities to the changing requirements of the client systems to which they were originally allocated: thus the capabilities, and not the mappings, are maintained and the mappings, not the capabilities, are dynamic, so as to optimise the allocation of client systems to cells. It also minimises the number of clients having to work to more than one datacentre, which leads to delays in processing as the datacentres need to communicate with each other. Configuration of the datacentres X, Y, Z themselves is required to set up the system, but subsequently only if a re-optimization of the existing mappings cannot satisfy the changed demand profiles, for example because the overall balance of services available from the datacentres no longer matches the services required.