Virtual Server Rack Workload Domains With Policy-Based Resource Reallocation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Data center operators face challenges in providing cost-effective cloud computing services by optimizing infrastructure resources, as existing techniques require detailed knowledge of hardware configurations and are complex to manage, especially when deploying and maintaining workload domains across multiple racks.
Innovation Solution
The method involves mapping workload domains to management clusters in a single rack deployment, allowing for easier configuration and management, enabling cross-rack cluster options, and providing users with intuitive interfaces to define security, capacity, availability, and performance requirements without needing in-depth knowledge of server rack hardware.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If detailed knowledge of hardware configurations is required for deploying workload domains, then precise control over resource allocation is achieved, but system complexity and difficulty of management increase
Solution Approach 1:
The patent introduces management clusters as an intermediary layer between users and physical hardware. Management clusters abstract the complex hardware configurations (rack units, servers, storage devices) into simplified logical groupings, allowing users to deploy workload domains without needing detailed hardware knowledge while maintaining precise resource control through the cluster management layer
2Adaptability or versatility
If workload domains are deployed across multiple racks, then resource capacity and scalability are improved, but deployment complexity and management difficulty increase
Solution Approach 1:
The patent segments the data center infrastructure into multiple management clusters, each corresponding to a logical rack or rack group. This segmentation allows workload domains to be distributed across multiple racks while maintaining simplified management within each cluster, reducing overall deployment complexity through modular organization
Solution Approach 2:
Management clusters are designed as universal containers that can hold diverse hardware resources (compute, storage, networking) from multiple racks. This multi-functionality allows the same management cluster interface and workload domain deployment mechanism to work across heterogeneous hardware configurations, simplifying cross-rack deployments
3Adaptability or versatility
If users need to configure security, capacity, availability, and performance requirements, then workload domain customization is improved, but configuration complexity increases
Solution Approach 1:
The patent allows users to configure workload domains by specifying parameter ranges and constraints (security levels, capacity requirements, availability SLAs, performance thresholds) rather than detailed hardware configurations. The management cluster layer automatically translates these high-level parameters into specific resource allocations, maintaining configuration flexibility while reducing complexity
Data Source
AI summary
Methods and apparatus to manage workload domains in virtual server racks are disclosed. An example apparatus includes a policy enforcer to evaluate whether capacities of the plurality of workload domains comply with policy capacity levels of respective user-defined policies for the plurality of workload domains, and a resource manager to, when a first workload domain has a first quantify of resources that exceeds a first policy capacity level of a first user-defined policy, identify a second workload domain requesting a second quantity of resources, when the second quantity of resources is equal to the first quantity of resources, allocate the first quantity of resources from the first workload domain to the second workload domain to satisfy the first policy capacity level, when the second quantity of resources is less than the first quantity of resources, allocate the second quantity of resources from the first workload domain to the second workload domain, and de-allocate remaining resources from the first workload domain to a shared resource pool to satisfy the first policy capacity level.