Zone Storage Scaling-In with Read-Only Replication Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional methods for scaling geographically diverse data storage systems, particularly during scaling-in, result in data accessibility issues and resource-intensive recovery processes, leading to incomplete or delayed data availability due to the asynchronous nature of chunk replication and the need for zone recovery.
Innovation Solution
Implementing a method that designates a zone to be removed as read-only, allowing for continued data access and replication of up-to-date data to other zones, thereby reducing resource consumption and ensuring data availability during the scaling-in process by using replication and merging techniques based on hashing functions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional scaling-in methods are used in geographically diverse data storage systems, then data can be recovered from replicated zones, but data accessibility is delayed and recovery processes are resource-intensive due to asynchronous chunk replication
Solution Approach 1:
The patent applies preliminary action by pre-replicating directory table data to a designated zone before the scaling-in operation begins. This ensures that when a zone is removed from the system, the directory table data is already available at the destination, eliminating the need for time-consuming recovery operations after the fact. The directory table is replicated in advance using hashing functions to determine the target zone, ensuring data is accessible immediately upon scaling-in.
2Reliability
If conventional scaling-in methods are used in geographically diverse data storage systems, then data can be recovered from replicated zones, but the recovery process consumes significant computational resources
Solution Approach 1:
The patent reduces computational resource consumption during scaling-in by performing directory table replication in advance, before the zone removal operation. This preliminary action shifts the computational burden from the critical scaling-in moment to a planned maintenance window, allowing resources to be used more efficiently when actually needed for data access rather than for emergency recovery operations.
Solution Approach 2:
The patent uses copying mechanisms to replicate directory table data from the source zone to the designated destination zone. By using hashing functions to determine replication targets and employing efficient copy operations, the system minimizes the computational resources required during the scaling-in process while ensuring data integrity and availability.
3Productivity
If directory table data is not replicated before zone removal, then scaling-in can proceed quickly, but data accessibility issues occur and data may be lost or become unavailable
Solution Approach 1:
The patent resolves this contradiction by implementing preliminary replication of the directory table data before the zone removal operation. This advance preparation ensures that data completeness is maintained while allowing the actual scaling-in operation to proceed quickly, as the destination zone already has the necessary directory table data ready for immediate use.
Data Source
AI summary
Preservation of data during scaling of a geographically diverse data storage system is disclosed. In regard to scaling-in, a first zone storage component (ZSC) can be placed in read-only (RO) mode to allow continued access to data stored on the first ZSC, completion of previously queued operations, updating of data chunks, etc. Data chunks can comprise metadata stored in directory table partitions organized in a tree data structure scheme. An updated data chunk of the first ZSC can be replicated at other ZSCs before deleting the first ZSC. A first hash function can be used to distribute portions of the updated data chunk among the other ZSCs. A second hash function can be used to distribute key data values corresponding to the distributed portions of the updated data chunk among the other ZSCs. Employing the first and second hash functions can result in more efficient use of storage space and more even distribution of key data values when compared to simple replication of a data chunk of the first ZSC by the other ZSCs.