Range-Lock Extent Tracking for IO Hot Spot Identification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional data storage systems struggle to distinguish between LUN slices with evenly distributed access and those with active and inactive data blocks due to coarse 1 GB granularity, leading to inefficient data placement and high overhead in tracking individual host IO operations at the data block level.
Innovation Solution
Monitoring range locks on extents within a data storage apparatus to identify IO hot spots, which allows for precise data placement by tracking locks on ranges of extents and updating a lock history database with lock type, storage object identifier, timestamp, and range data, enabling efficient identification of active extents without excessive system resource usage.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If data placement operates at LUN slice granularity (1 GB), then system complexity is reduced and processing is simplified, but measurement precision deteriorates because the system cannot distinguish between evenly distributed access patterns and concentrated hot spots
Solution Approach 1:
The patent segments the LUN slice into smaller units called extents (typically 64 KB or 128 KB) for more granular tracking. Instead of treating the entire 1 GB LUN slice as a single unit, the system divides it into multiple extents that can be individually tracked for access patterns, enabling differentiation between hot spots and evenly distributed access while maintaining manageable system complexity
Solution Approach 2:
The patent introduces range locks as an intermediary mechanism between the host IO operations and the access tracking system. Range locks capture information about which extents were accessed during host operations, serving as a mediator that bridges the gap between coarse LUN slice tracking and fine-grained extent-level analysis without requiring direct implementation of complex block-level counting
2Measurement precision
If the system tracks individual host IO operations at the data block level, then measurement precision improves for identifying hot spots, but device complexity and resource consumption increase prohibitively
Solution Approach 1:
The patent extracts only the necessary information from individual host IO operations - specifically the range locks that indicate which extents were accessed. Instead of tracking every detail of block-level operations, the system extracts and stores only the extent range information in the lock history database, significantly reducing processing overhead while maintaining sufficient precision for hot spot identification
Solution Approach 2:
The patent changes the tracking parameter from individual block-level IO counts to range-lock-based extent access patterns. By using range locks that capture extent ranges rather than counting individual blocks, the system achieves better measurement precision for hot spots while reducing the volume of data to process and store
3Ease of operation
If the system uses LUN slice granularity for data placement, then ease of operation is improved, but reliability deteriorates because data placement decisions cannot account for uneven access distributions within slices
Solution Approach 1:
The patent applies local quality by enabling data placement decisions to be made at the extent level rather than uniformly across the entire LUN slice. The system identifies which specific extents within a slice are hot spots based on access patterns, allowing data placement to treat different regions of the slice differently - moving hot data to faster storage tiers while leaving cold data in slower tiers, thereby improving placement effectiveness
Data Source
AI summary
A technique of identifying IO hot spots is performed in a data storage apparatus (e.g., a file server). The technique involves updating, in response to host IO operations which lock ranges of extents prior to accessing the ranges of extents, contents of a lock history database based on the ranges of extents which were locked by the host IO operations. The technique further involves receiving a lock history request. The technique further involves providing, in response to the lock history request, the contents of the lock history database to identify, as the IO hot spots, extents which were locked by the host IO operations.