Volume Partition Placement for Storage Reliability
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Determining optimal distribution of volume partitions across data storage servers for remote computing services to achieve performance, durability, and availability is challenging due to changing conditions and constraints over time.
Innovation Solution
A system and method that dynamically determine the optimal distribution of volume partitions based on factors like available space, I/O operations, replica pair counts, and network locality, using a server suitability score to recommend placement across multiple servers and racks, ensuring rack diversity and periodic rebalancing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If volume partitions are distributed across multiple servers to improve durability and availability, then reliability improves, but device complexity increases
Solution Approach 1:
The volume is divided into multiple partitions that can be independently distributed across different servers. Each partition can be placed on different servers to achieve fault tolerance, allowing the system to segment data and distribute it for improved reliability without requiring complete system complexity management.
Solution Approach 2:
The placement of volume partitions is made dynamic rather than static. The system can periodically rebalance partitions based on changing conditions such as server availability, storage capacity, and performance metrics, allowing the distribution to adapt over time while maintaining reliability.
2Adaptability or versatility
If volume partitions are dynamically redistributed to adapt to changing conditions, then adaptability improves, but loss of time increases
Solution Approach 1:
Instead of continuous redistribution, the system performs periodic rebalancing at scheduled intervals or in response to specific events. This allows the system to adapt to changing conditions without the time cost of constant redistribution, achieving adaptability while minimizing time loss through timely but not continuous actions.
3Reliability
If replica pairs are placed on different racks to reduce failure risk, then reliability improves, but device complexity increases
Solution Approach 1:
The system applies different placement strategies for different replica pairs based on local conditions. Not all replica pairs need to be distributed across different racks, but rather the placement is optimized locally for each partition considering factors like network locality, storage capacity, and failure risk, achieving reliability without uniform complexity across the entire system.
Data Source
AI summary
A system and method for providing sets of partition placements, the system and method including, receiving at least one placement request for a set of partitions of a volume. Based at least in part on counts of pairs of partitions hosted by pairs of computing devices, the system and method further includes determining how suitable the pairs of computing devices are for placement of partitions of the set of partitions, generating a set of placements based at least in part on the determination, and providing the set of placements in response to the at least one placement request.