Remote Storage Cache Bypassing Replication Node for IO Response
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Remote replication in storage systems increases latency and bandwidth usage due to the additional network hops required for data consistency checks and response routing, which can significantly impact storage system throughput and bandwidth.
Innovation Solution
Implementing replication logic that allows direct communication paths between compute nodes and storage nodes, bypassing the replication node for response data, thereby reducing the number of network hops and improving bandwidth utilization.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If remote replication functionality is implemented to maintain back-up data in a remote location, then data reliability is improved, but latency associated with servicing input-output operations increases
Solution Approach 1:
The patent segments the data path by introducing a cache layer at the remote location that can serve read requests independently. This allows read operations to be fulfilled from the cache without involving the primary storage system, thereby reducing latency while maintaining data reliability through the cached copies.
Solution Approach 2:
The patent implements preliminary action by pre-caching replicated data at the remote location before it is needed for read operations. This advance preparation allows the system to serve read requests from the cache immediately, reducing the latency that would otherwise occur when data must be retrieved from the primary storage system.
2Reliability
If remote replication functionality is implemented to maintain back-up data in a remote location, then data reliability is improved, but bandwidth usage of the storage system increases
Solution Approach 1:
The patent applies local quality by enabling the remote cache to independently serve read requests locally without requiring continuous communication with the primary storage system. This local化处理 of read operations significantly reduces network bandwidth consumption while maintaining data reliability through the pre-cached copies.
Solution Approach 2:
The patent ensures continuity of useful action by maintaining the cached data at the remote location, allowing read operations to continue being served from the cache without interrupting the primary storage system's bandwidth. This continuous local serving capability reduces overall bandwidth usage while preserving data accessibility.
3Loss of time
If direct communication paths between compute nodes and storage nodes are implemented, then latency is reduced, but device complexity increases
Solution Approach 1:
The patent introduces a cache as an intermediary component at the remote location that mediates between compute nodes and storage nodes. This intermediary enables direct communication paths for read operations, reducing latency, while managing the complexity by handling cache maintenance and consistency automatically.
Solution Approach 2:
The patent implements multi-functionality by designing the remote cache to serve multiple purposes: it acts as a buffer for replicated data, a serving mechanism for read requests, and a complexity-absorbing layer that handles cache management. This universal component reduces latency across different operation types while consolidating complexity into a single manageable structure.
Data Source
AI summary
An apparatus comprising a processing device is configured to communicate with a storage node of a storage system via at least one network and to communicate with a compute node that implements at least a portion of an application via the at least one network. The at least one processing device is configured to obtain an input-output request associated with the application from the compute node via the at least one network and to associate information corresponding to the compute node with the input-output request. The at least one processing device is further configured to submit the input-output request and the associated information that corresponds to the compute node to the storage node. The storage node is configured to submit a response to the input-output request to the compute node via the at least one network based at least in part on the information.