Overlay Network Hash Segmentation for Distributed Server Load
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In distributed systems, especially those with a large number of servers, maintaining network efficiency and reducing communication loads become challenging due to high destination information and connection resource requirements, leading to increased maintenance costs and latency.
Innovation Solution
The proposed solution involves constructing an overlay network using a split hash method, where the hash space is divided into areas, and parent servers are selected to manage access within these areas, reducing the load on the network by allowing dynamic configuration changes and localized management of server additions and withdrawals.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If a distributed system uses a large number of servers to increase processing capacity, then system productivity is improved, but network maintenance load and communication overhead increase
Solution Approach 1:
The patent segments the distributed system into hierarchical groups and subgroups, where each group manages a subset of servers. This segmentation reduces the network maintenance load by localizing management responsibilities to group management servers rather than requiring global coordination across all servers, thus resolving the contradiction between high processing capacity and reduced maintenance complexity
Solution Approach 2:
The patent introduces group management servers as intermediaries between clients and individual servers. These intermediaries handle destination information management and request routing, reducing the communication overhead on the network by consolidating management functions at the group level rather than requiring direct client-to-server communication for all operations
2Device complexity
If destination information is managed centrally to simplify system structure, then device complexity is reduced, but system reliability deteriorates due to single point of failure
Solution Approach 1:
The patent divides the centralized destination information management into distributed group management servers, each responsible for specific groups of servers. This segmentation eliminates the single point of failure while maintaining simplified structure within each group, as each group management server independently manages its own subset without requiring a single global authority
Solution Approach 2:
The patent implements local quality by allowing each group management server to independently manage destination information for its specific group. This local autonomy ensures that failures in one group do not affect other groups, improving overall system reliability while maintaining simple localized management structures
3Speed
If all servers are connected in a fully meshed network to reduce communication hops, then response speed is improved, but network maintenance load increases significantly
Solution Approach 1:
The patent segments the fully meshed network into hierarchical groups and subgroups, where connection resources are localized within groups rather than requiring global mesh connections. This segmentation reduces the number of required connections while maintaining fast response times by limiting the scope of communication to local groups first, then escalating to higher levels only when necessary
4Adaptability or versatility
If dynamic server addition and deletion is supported to increase system adaptability, then adaptability is improved, but network maintenance load increases
Solution Approach 1:
The patent introduces group management servers as intermediaries that handle dynamic server addition and deletion operations. When servers are added or removed, only the relevant group management server needs to update its destination information, rather than requiring system-wide updates. This intermediary approach maintains high adaptability while reducing maintenance costs by localizing the impact of configuration changes
Data Source
AI summary
An information processing device having any of hash values associated with a hash space based on distribution hash information. The device acquires access information to an information processing device as a target, the access information being transmitted from another information processing device on the basis of first distribution hash information using the hash value in the hash space between first devices belonging to one of groups divided according to a sequence in terms of the magnitudes of the hash values associated with the hash space, stores second distribution hash information using the hash value in the hash space pertaining to a second device belonging to the group to which the first device belongs, searches second distribution hash information for a second device corresponding to the hash value generated from the access information, and transmits access information to the retrieved second device by the searching.