Cloud Storage Allocation via Workload Profiling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current cloud computing environments inefficiently allocate block-based data storage due to reliance on low-level metrics like capacity and IOPS, leading to under or over provisioning as applications have more complex storage requirements that are not accurately met by existing allocation methods.
Innovation Solution
A method and system for dynamically allocating cloud storage based on specific storage attributes such as read/write ratio, sequential/random IO, and IO size, which configures infrastructure to optimize storage resources without requiring application owners to assemble their own infrastructure, using a profiling system that adjusts to changing workload characteristics.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If storage is allocated based on low-level metrics (capacity and IOPS), then allocation simplicity is maintained, but storage requirement accuracy deteriorates
Solution Approach 1:
The patent transitions from low-level storage metrics (capacity, IOPS) to high-level application characteristics (read-write ratio, sequential-random IO pattern, IO size) to describe storage requirements. This parameter transformation enables accurate matching of storage workloads to infrastructure without increasing allocation complexity for end users.
2Measurement precision
If application owners assemble their own storage infrastructure, then storage requirement accuracy improves, but device complexity increases
Solution Approach 1:
The patent implements automated workload analysis that profiles application storage characteristics and automatically configures appropriate storage infrastructure. This eliminates the need for application owners to manually assemble storage components while maintaining accurate requirement matching through systematic workload characterization.
Solution Approach 2:
The patent introduces an intermediary storage profiling system between the application and infrastructure layers. This mediator analyzes workload characteristics, determines storage requirements, and configures infrastructure automatically, bridging the gap between simple application requests and complex storage configuration needs.
3Reliability
If overprovisioning is used to avoid reprovisioning, then reliability improves, but resource efficiency deteriorates
Solution Approach 1:
The patent implements continuous monitoring and profiling of actual storage workload characteristics, using this feedback to dynamically adjust storage allocation. This enables the system to provision storage based on actual needs rather than estimates, maintaining reliability while eliminating excess resource allocation.
Data Source
AI summary
A method of allocating cloud storage based on storage profiles includes receiving a storage request from a virtual machine associated with a tenant at a computing apparatus having connectivity to a network. The storage request is analyzed to determine one or more storage attributes of the storage request. An infrastructure is configured in a configuration based on the one or more storage attributes and data associated with the storage request is stored in the infrastructure in accordance with the configuration.