Anomaly Detection in Shared IO Resources via Dynamic Window Adjustment
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In modern data centers, unmanaged clients that are not bound by Quality of Service (QoS) policies can present large workloads to shared IO resources, disrupting the resource allocation and bandwidth management for clients governed by QoS policies, leading to anomalies that impact IO access bandwidth and service distribution.
Innovation Solution
A method is implemented to detect anomalies in shared IO resources by monitoring IO statistics such as latency and window sizes, and when an anomaly is detected, the window sizes of managed hosts are increased to their maximum or proportionally adjusted based on QoS shares to mitigate the impact.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If unmanaged clients are allowed to access shared IO resources without QoS policies, then ease of operation and system flexibility are improved, but IO access bandwidth and resource distribution are disrupted
Solution Approach 1:
The patent introduces an intermediary mechanism that monitors IO statistics (latency, window sizes) and detects anomalies caused by unmanaged clients. When anomalies are detected, the system automatically adjusts window sizes of managed hosts to compensate, thereby protecting IO access bandwidth management while maintaining open access for unmanaged clients
Solution Approach 2:
The system continuously monitors IO statistics and uses this feedback to detect anomalies in real-time. Based on the feedback from anomaly detection, the system dynamically adjusts window sizes to maintain fair resource distribution, creating a closed-loop control mechanism that resolves the contradiction between open access and bandwidth management
2Reliability
If QoS policies are enforced to manage IO bandwidth, then IO access bandwidth management is improved, but device complexity and policy management overhead increase
Solution Approach 1:
The patent enables managed hosts to self-adjust their window sizes automatically when anomalies are detected. Instead of requiring complex centralized QoS policy management, each host independently monitors its own IO statistics and adjusts its parameters, reducing management overhead while maintaining bandwidth control
Solution Approach 2:
The system manages IO bandwidth by dynamically changing the window size parameter of hosts rather than implementing complex QoS policies. This simple parameter adjustment mechanism provides effective bandwidth management with minimal device complexity
3Reliability
If window sizes are increased to maximum to protect against anomalies, then IO access bandwidth for managed hosts is protected, but resource distribution fairness among multiple managed hosts may be compromised
Solution Approach 1:
The patent applies different window size adjustment strategies to different hosts based on their QoS shares. When anomalies are detected, hosts with higher QoS shares receive larger window size increases, ensuring that resource distribution fairness is maintained while still protecting IO access bandwidth for all managed hosts
Data Source
AI summary
An anomaly in a shared input/output (IO) resource that is accessed by a plurality hosts or clients is detected when a host that is not bound by any QoS policy presents large workloads to a shared IO resource that is also accessed by hosts or clients that are governed by QoS policy. The anomaly detection triggers a response from the hosts or clients as a way to protect against the effect of the anomaly. The response is an increase in window sizes. The window sizes of the hosts or clients may be increased to the maximum window size or in proportion to their QoS shares.