Multi-tenant Container Scheduler for Cloud Resource Allocation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Cloud-based big data services face challenges in scalability and performance due to the need for large-scale compute and storage resources, which can be costly and cumbersome to manage, especially for organizations without the resources to implement and maintain their own infrastructure.
Innovation Solution
A container-based, multi-tenant computing infrastructure that dynamically allocates compute resources among clusters based on demand, using a scheduler to move underutilized containers to areas with pending resources, optimizing resource utilization and minimizing costs through automated recycling and health monitoring of nodes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of energy
If cloud-based big data services are used to reduce costs and improve resource utilization, then resource efficiency and cost savings are improved, but scalability and performance deteriorate due to resource contention and shared infrastructure
Solution Approach 1:
The patent segments the cloud infrastructure into multiple isolated clusters, each dedicated to specific tenants. This segmentation prevents resource contention between tenants while maintaining cloud-based resource efficiency. Each cluster can be independently scaled and managed, resolving the contradiction between shared resource efficiency and individual tenant scalability needs.
Solution Approach 2:
The patent implements dynamic resource allocation within clusters through container orchestration and automated scaling mechanisms. Resources can be dynamically provisioned and deprovisioned based on demand, allowing the system to maintain high resource utilization while providing scalable capacity to individual tenants as needed.
2Productivity
If large-scale compute and storage resources are procured for on-premises infrastructure, then performance and control are improved, but cost and operational complexity worsen
Solution Approach 1:
The patent introduces a cluster management platform as an intermediary layer between tenants and the underlying infrastructure. This platform handles provisioning, monitoring, scaling, and maintenance operations automatically, allowing tenants to access high-performance resources without bearing the operational complexity of managing large-scale infrastructure directly.
Solution Approach 2:
The system implements self-service capabilities through automated resource provisioning and management. The cluster management platform automatically handles resource allocation, scaling, and maintenance tasks without requiring manual intervention from tenants, thereby providing high-performance resources while minimizing operational complexity for users.
3Loss of energy
If cloud vendors share support among many clients to reduce costs, then resource utilization and cost savings are improved, but service quality and performance worsen due to resource contention
Solution Approach 1:
The patent segments the multi-tenant environment into isolated clusters, where each cluster provides dedicated resources to specific tenants. This segmentation ensures that resource utilization benefits of cloud sharing are maintained while preventing resource contention from degrading service quality. Each tenant experiences consistent performance as if having dedicated resources.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Techniques are described for managing compute and storage resources in a container-based, multi-tenant, computing infrastructure. In some implementations, a plurality of containers may each execute an instance of a software framework. Each container may include compute capacity and storage capacity according to the executed framework. Containers may be allocated among clusters in a multi-tenant, cloud-based computing environment based on measured demand for compute capacity and/or storage capacity in the various clusters. In some implementations, a scheduler (e.g., a Level 2 scheduler, or a scheduler of schedulers) determines whether to move compute containers between clusters within a multi-tenant environment, to provide performance improvements, resource utilization efficiency, and cost reduction within the data center, and to satisfy tenant requests for compute and/or storage resources.