Virtual Block Storage Appliances for Multi-Site Cache Synchronization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional multi-site storage systems are limited to active-passive data access, where only the primary file system is read-write, while replicated systems are read-only, leading to prolonged file system recovery times due to inconsistent states between disk and cache resources.
Innovation Solution
Implementing virtual block storage appliances for both disk-level and cache-level synchronization in a multi-site storage system, allowing both primary and replicated file systems to be mounted as read-write, ensuring consistent states and enabling active-active data access.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If replicated file systems are mounted as read-only to ensure consistency, then data consistency is improved, but data access functionality deteriorates
Solution Approach 1:
The patent segments the storage system into cache storage and disk storage components, allowing different access modes for each. The cache storage is mounted as read-write while the disk storage maintains read-only access, enabling simultaneous data access functionality and consistency. This segmentation resolves the contradiction by allowing the replicated file system to be mounted as read-write for cache while maintaining read-only access for disk, thus improving both data access functionality and data consistency.
2Reliability
If replicated file systems are mounted as read-only, then data consistency is maintained, but file system recovery time increases
Solution Approach 1:
The patent implements preliminary action by maintaining synchronized cache storage at the remote site before failure occurs. The cache storage is kept updated with recent data changes, so when a failure happens, the remote site can immediately activate the synchronized cache storage as read-write, dramatically reducing recovery time. This preliminary synchronization of cache storage eliminates the need for lengthy recovery processes while maintaining data consistency.
3Device complexity
If only disk replication is implemented, then system complexity is reduced, but data consistency deteriorates
Solution Approach 1:
The patent merges disk-level replication and cache-level replication into a unified multi-site storage system. By combining both replication mechanisms, the system achieves comprehensive data consistency across sites. The cache-level synchronization complements the disk-level replication, ensuring that both persistent storage and volatile cache remain consistent, thereby improving data consistency without excessively increasing system complexity.
Data Source
AI summary
One or more processing platforms are configured to implement at least a first site of a multi-site storage system. The first site comprises a first file system resident at the first site, and a replicated version of a second file system. The second file system is resident at a second site of the multi-site storage system, and the replicated version of the second file system is resident at the first site. Virtual block storage appliances running on respective virtual machines of the first site interact with one another and with similar appliances at the second site to allow the replicated version of the second file system to be synchronized at a cache level with the second file system, and to allow a replicated version of the first file system at the second site to be synchronized at a cache level with the first file system at the first site.