Zone Management for Compute-Centric Object Stores
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Cloud computing systems face inefficiencies due to the use of virtual machines (VMs) that require hardware emulation, leading to reduced performance and instability as the number of VMs increases, especially in managing varying workloads and ensuring fair resource allocation in multitenant environments.
Innovation Solution
The system employs a reserve zone pool and a non-reserve zone pool of virtual operating system containers, dynamically assigning them to task groups based on expected and actual concurrency rates to optimize resource allocation and prevent instability, allowing for efficient execution of compute operations directly on object stores without the need for complex hypervisor management.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If virtual machines are used to manage cloud workloads, then resource isolation and multitenancy are achieved, but hardware emulation reduces performance and increases system complexity
Solution Approach 1:
The patent extracts the hardware emulation layer from the virtualization architecture by using operating system containers instead of full virtual machines. This removes the hypervisor and hardware emulation components while preserving the essential function of resource isolation through container-based virtualization, thereby reducing system complexity without sacrificing reliability.
Solution Approach 2:
The patent creates a simplified copy of the virtual machine concept using operating system containers that replicate the resource isolation and multitenancy capabilities without the overhead of hardware emulation. These containers provide a lightweight alternative that maintains the functional benefits while eliminating the performance penalties of full VM emulation.
2Adaptability or versatility
If the number of virtual machines increases to handle varying workloads, then resource flexibility improves, but system instability increases due to hypervisor coordination overhead
Solution Approach 1:
The patent implements dynamic workload management through operating system containers that can be rapidly created, moved, and destroyed without the coordination overhead of hypervisor-managed VMs. The container-based approach allows flexible adaptation to varying workloads while maintaining system stability through simpler resource management and reduced inter-VM coordination requirements.
3Reliability
If hardware emulation is used in virtual machines, then complete system platform isolation is achieved, but execution performance deteriorates
Solution Approach 1:
The patent replaces the mechanical hardware emulation system with a software-based container isolation mechanism. Instead of emulating hardware through software layers, the system uses operating system-level containerization to achieve platform isolation, eliminating the performance overhead of hardware emulation while maintaining the essential security and isolation benefits.
4Ease of operation
If virtual machines are used for compute operations, then resource allocation is simplified, but resource utilization efficiency decreases due to emulation overhead
Solution Approach 1:
The patent employs lightweight, short-lived operating system containers instead of heavy, persistent virtual machines. These containers can be rapidly instantiated and terminated with minimal overhead, providing efficient resource allocation and high utilization rates. The disposable nature of these containers allows for rapid scaling and adaptation to workload demands without the performance penalty of hardware emulation.
Data Source
AI summary
Zone management of compute-based object stores is provided herein. An exemplary method may include assigning a virtual operating system container from the reserve zone pool to a task group, the task group including a set of tasks for a phase of a first request, and executing the set of tasks within the assigned virtual operating system container.