Transactional I/O Scheduling for SSD Backup Performance
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current deduplication backup systems face limitations in optimally applying Quality of Service (QoS) agreements to specific backup applications due to lack of device-type awareness, leading to suboptimal performance, especially in environments where storage is a bottleneck.
Innovation Solution
Implementing an I/O tagging module and transactional I/O scheduler that tags sibling I/Os initiated from a parent backup application with a unique tag ID, allowing the scheduler to prioritize and dispatch them based on time of arrival and QoS policies, thereby optimizing I/O operations across different storage devices, including SSD and HDD.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If QoS parameters are applied at the system level without device-type awareness, then the system can maintain simple QoS management, but the performance guarantee for backup applications deteriorates due to inability to prioritize I/Os on specific storage devices
Solution Approach 1:
The patent applies local quality by differentiating QoS treatment based on storage device type (SSD vs HDD). The system identifies I/O operations targeting specific device types and applies device-aware scheduling policies, ensuring that backup application I/Os receive prioritized treatment on appropriate storage devices while internal maintenance operations are throttled. This localized differentiation enables performance guarantees without requiring complete system-wide complexity.
2Productivity
If I/O operations from backup applications and internal maintenance operations share the same storage devices, then storage resources are utilized efficiently, but the QoS agreement for backup applications cannot be honored due to interleaving with internal operations
Solution Approach 1:
The patent segments I/O operations into distinct categories: backup application I/Os and internal maintenance operations. By tagging and identifying these segments separately, the system can apply different scheduling policies to each segment. This segmentation allows the system to maintain storage resource utilization efficiency while ensuring that backup application I/Os receive guaranteed QoS treatment without interleaving with internal operations.
Solution Approach 2:
The system performs preliminary action by pre-identifying and tagging I/O operations as belonging to backup applications or internal maintenance operations before they are dispatched to storage devices. This preliminary classification enables the scheduler to prepare appropriate QoS treatments in advance, ensuring that backup application I/Os are prioritized and scheduled without interleaving with internal operations, thus honoring QoS agreements while maintaining efficient resource utilization.
3Productivity
If the I/O Scheduler prioritizes backup application workload over internal management operations on a single disk, then backup performance is improved, but the system cannot optimally manage I/Os across multiple storage devices with different types
Solution Approach 1:
The patent implements universality by creating a device-aware QoS framework that can operate across multiple storage device types (SSD and HDD). The system uses a unified tagging mechanism to identify backup application I/Os regardless of which storage device they target, and applies consistent QoS policies across the entire storage subsystem. This universal approach enables the system to maintain backup application performance prioritization while adapting to the specific characteristics of different storage devices, making the QoS agreement applicable across the entire multi-device environment.
Data Source
AI summary
Embodiments are described for prioritizing input/output (I/O) operations dispatched to a solid-state device (SSD) cache in a network, by defining a maximum write I/O operation size for writing data to the SSD cache, splitting large write I/O operations into smaller write I/O operations, each with a size less than the maximum write I/O operation size, interleaving cache read I/O operations in between the smaller write I/O operations, and performing the cache read I/O operations and the smaller write I/O operations in an order created by the interleaving. The network may comprise a deduplication backup system storing data to storage media including the SSD cache.