Index Deletion via Contiguous Storage Unit Batching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for deleting indexes in an internal memory require multiple IO operations, leading to inefficient storage reads and impaired system performance due to the need to read and delete indexes corresponding to multiple data blocks, which affects overall system performance when cache space is insufficient.
Innovation Solution
The method involves selecting a target storage unit with consecutively stored indexes, reading all indexes at once using a start address and length, deleting them from the index table, and marking the unit as empty, thereby reducing the number of IO operations required for index deletion.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If indexes corresponding to multiple data blocks are read and deleted separately, then each index can be deleted individually, but multiple IO operations are required which reduces system performance
Solution Approach 1:
The patent merges multiple discrete index deletion operations into a single batch operation. By organizing indexes into contiguous ranges and processing them together through one IO operation, the system eliminates the overhead of multiple separate read-delete cycles, directly resolving the contradiction between individual index deletion capability and system performance.
Solution Approach 2:
The patent performs preliminary organization of indexes into contiguous ranges before the deletion operation. By pre-grouping indexes that will be deleted together and ensuring they are stored contiguously, the system prepares the data structure in advance to enable single-IO batch deletion, avoiding the need for multiple sequential operations.
2Reliability
If multiple IO operations are performed to read indexes for deletion, then all necessary indexes can be accessed, but storage read operations increase which impairs system performance
Solution Approach 1:
The patent combines multiple storage read operations into a single read operation by identifying contiguous index ranges. This merging approach ensures that all necessary indexes are accessed (maintaining reliability) while reducing the total number of IO operations from multiple separate reads to one consolidated read, thereby improving system performance.
3Ease of operation
If indexes are stored discretely in storage units, then each index can be accessed independently, but reading all indexes requires multiple IO operations
Solution Approach 1:
The patent segments the storage structure into units where indexes are organized in contiguous ranges. This segmentation allows indexes to be accessed individually when needed while also enabling efficient batch reading of entire index ranges through single IO operations, resolving the contradiction between individual access flexibility and batch reading efficiency.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A method for deleting indexes in an internal memory is disclosed. The method includes: selecting a to-be-evicted target storage unit from the plurality of storage units; reading all indexes in the target storage unit, where all the indexes in the target storage unit are consecutively stored in the target storage unit; deleting all the read indexes from an index table in the internal memory; and marking the target storage unit as empty. In this technical solution, indexes corresponding to a plurality of pieces of data cached in a storage can be read through one IO, so that the same indexes can be deleted from the index table in the internal memory.