Metadata Cache Coherence in Clustered File Systems
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current clustered SAN-based file systems face complexity and performance bottlenecks in ensuring metadata cache coherence across multiple nodes, often requiring complex networking setups and asymmetric implementations that designate specific nodes as metadata servers, leading to potential bottlenecks.
Innovation Solution
Implementing a method using SCSI command extensions and storage system information to ensure metadata cache coherence by comparing version numbers, designating lock managers, and utilizing barriers to synchronize metadata modifications across multiple hosts accessing shared files.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If complex networking setups are used to ensure metadata cache coherence, then cache coherence can be maintained, but device complexity and implementation difficulty increase
Solution Approach 1:
The patent extracts the cache coherence management function from the network layer and relocates it to the storage device itself. The storage device maintains version numbers and lock information locally, eliminating the need for complex network-based coordination protocols while ensuring metadata cache coherence across multiple nodes.
Solution Approach 2:
The storage device performs self-service by autonomously managing metadata coherence through internal version tracking and lock mechanisms. Each node independently monitors version numbers stored in the storage device, and the storage device itself handles the coordination of metadata updates without requiring external network-based lock managers.
2Ease of operation
If asymmetric file system implementation with metadata servers is used, then metadata management can be centralized, but performance bottlenecks and availability issues arise
Solution Approach 1:
The patent segments the metadata management function across all nodes rather than concentrating it in dedicated metadata servers. Each node maintains its own copy of version numbers and lock information, allowing parallel and distributed metadata operations that eliminate single-point bottlenecks while maintaining operational simplicity through local caching.
Solution Approach 2:
The patent creates equipotentiality by allowing any node to serve as a metadata server for any file. Rather than having designated metadata servers with specialized roles, all nodes have equal capability to manage metadata for any shared file, distributing the load and eliminating performance bottlenecks associated with centralized metadata management.
3Reliability
If version number comparison and lock management are implemented, then metadata cache coherence is ensured, but processing time increases
Solution Approach 1:
The patent applies preliminary action by pre-storing version numbers and lock information in the storage device before metadata changes occur. Nodes read these version numbers locally rather than querying them during operations, and the storage device pre-manages lock states, reducing processing time while maintaining coherence through advance preparation of coherence data.
Data Source
AI summary
Metadata of a shared file in a clustered file system is changed in a way that ensures cache coherence amongst servers that can simultaneously access the shared file. Before a server changes the metadata of the shared file, it waits until no other server is attempting to access the shared file, and all I/O operations to the shared file are blocked. After writing the metadata changes to the shared file, local caches of the other servers are updated, as needed, and I/O operations to the shared file are unblocked.