Symmetric Storage Allocation for Heterogeneous Distributed Nodes
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In distributed storage systems with heterogeneous access probabilities, existing methods struggle to optimize storage allocation to maximize the probability of successful data recovery, especially when nodes have varying reliabilities, leading to complex combinatorial optimization problems and inefficiencies.
Innovation Solution
The implementation of symmetric allocation techniques, including one-level, two-level, and k-level allocations, where the storage budget is spread evenly across subsets of nodes with different reliability levels, allowing for efficient allocation and improved recovery probabilities, using algorithms that simplify the optimization process and outperform existing methods.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If existing methods (large deviation inequalities and convex optimization) are used for storage allocation, then the optimization process becomes complex and computationally intensive, but the recovery probability may not be maximized efficiently
Solution Approach 1:
The patent segments the storage allocation problem into discrete levels (one-level, two-level, k-level symmetric allocations) where storage budget is distributed across hierarchical subsets of nodes. This segmentation transforms the complex continuous optimization problem into manageable discrete allocation scenarios, reducing computational complexity while maintaining high recovery probability through systematic distribution across multiple reliability levels.
Solution Approach 2:
The patent applies local quality by allocating different storage amounts to different subsets of nodes based on their individual reliability characteristics. Nodes with higher access probabilities receive different allocation weights compared to nodes with lower access probabilities, optimizing the recovery probability for each local subset while maintaining overall system efficiency.
2Productivity
If symmetric allocation is used to simplify the optimization process, then computational efficiency improves, but the allocation may not be optimal for all heterogeneous node configurations
Solution Approach 1:
The patent introduces dynamic allocation strategies where the symmetric allocation framework adapts to different node configurations through multiple levels. The k-level symmetric allocation dynamically adjusts the hierarchy and distribution based on the specific access probability profile of nodes, allowing the system to optimize recovery probability across diverse heterogeneous configurations while maintaining computational efficiency through structured approaches.
3Device complexity
If one-level symmetric allocation is used, then the allocation process is simple and efficient, but the recovery probability is lower compared to multi-level allocations
Solution Approach 1:
The patent extends the allocation problem from a single level to multiple levels, adding a hierarchical dimension to the allocation structure. This dimensional change allows the system to capture more nuanced reliability patterns by distributing storage across multiple subsets with different access probability characteristics, thereby increasing recovery probability while maintaining manageable complexity through structured hierarchy.
Data Source
AI summary
Allocation of storage budget in a computer-based distributed storage system is described, where associated computer-based storage nodes have heterogeneous access probabilities. The problem is to allocate a given storage budget across the available computer-based nodes so as to store a unit-size data object (e.g. file) with a higher reliability (e.g. increased probability for the storage budget to be recovered). Efficient algorithms for optimizing over one or more classes of allocations are presented. A basic one-level symmetric allocation is presented, where the storage budget is spread evenly over an appropriately chosen subset of nodes. Furthermore, a two-level symmetric allocation is presented, where the budget is divided into two parts, each spread evenly over a different subset of computer-based storage nodes, such that the amount allocated to each node in the first subset is twice that of the second subset. Further expansion of the two-level symmetric allocation is provided with a three-level and a generic k-level symmetric allocation.