Storage Volume Rebalancing via I/O Latency Monitoring
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Cloud computing systems face inefficiencies in managing storage resources, particularly in allocating and balancing storage volumes across storage controllers to maintain optimal performance, as existing methods do not effectively monitor and manage I/O latency, network latency, and network bandwidth utilization.
Innovation Solution
A method that monitors I/O latency, network latency, and network bandwidth utilization for each storage controller, determining thresholds and rebalancing storage volume distribution when these values exceed set limits, ensuring efficient allocation and balancing of storage resources by selecting appropriate storage controllers for new volume allocation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If storage volumes are allocated to storage controllers based on simple availability checks, then allocation speed is fast, but I/O latency performance deteriorates due to imbalanced distribution
Solution Approach 1:
The system continuously monitors I/O latency values for each storage controller and uses this feedback to dynamically adjust storage volume distribution. When I/O latency exceeds a threshold, the system triggers rebalancing operations to redistribute volumes, ensuring performance requirements are maintained while adapting to changing system conditions.
Solution Approach 2:
The storage allocation system transitions from a static allocation approach to a dynamic one where storage volume distribution is continuously adjusted based on real-time I/O latency measurements. The system can automatically rebalance storage volumes across controllers to optimize performance, making the allocation adaptive rather than fixed.
2Productivity
If storage volumes are evenly distributed across all storage controllers, then load balancing is improved, but I/O latency increases due to network bandwidth constraints
Solution Approach 1:
Instead of applying a uniform distribution policy across all storage controllers, the system implements local quality by allocating storage volumes based on the specific characteristics and current state of each controller. Controllers with better network bandwidth availability and lower I/O latency receive more storage volumes, while others receive fewer, creating a non-uniform but optimized distribution.
Solution Approach 2:
The system changes the allocation parameters dynamically based on monitored performance metrics. Rather than using a fixed even distribution ratio, the allocation decisions are adjusted according to real-time I/O latency values and network bandwidth utilization, allowing the system to optimize for current conditions rather than maintaining a static balance.
3Reliability
If storage volume rebalancing is performed frequently to maintain optimal performance, then I/O latency is minimized, but system complexity and overhead increase
Solution Approach 1:
The system implements periodic monitoring of I/O latency values and triggers rebalancing operations only when performance thresholds are exceeded, rather than continuously rebalancing. This periodic approach reduces the complexity and overhead of the rebalancing mechanism while still maintaining acceptable performance levels by acting only when necessary.
4Reliability
If monitoring and rebalancing operations are added to manage storage resources, then I/O latency performance is improved, but computational overhead increases
Solution Approach 1:
The storage management system performs self-service by automatically monitoring its own performance metrics and triggering rebalancing operations when needed, without requiring external intervention. This self-managed approach improves I/O latency performance through continuous optimization while minimizing unnecessary computational overhead by only performing monitoring and rebalancing when performance degradation is detected.
Data Source
AI summary
A method and technique for allocation and balancing of storage resources includes monitoring, for each of a plurality of storage controllers, an input/output (I/O) latency value based on an I/O latency associated with each storage volume controlled by a respective storage controller. An I/O latency value threshold is determined. Responsive to a change to the I/O latency value exceeding a threshold, storage volume distribution among the storage controllers is rebalanced.