Storage Grid with Channel Director for Low Latency IO
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Networked storage systems experience increased latency due to IO operations being performed over networks, which can be exacerbated by slow storage tiers and network protocols, affecting the performance of host devices and applications.
Innovation Solution
Integrating computation resources with a storage array through a storage grid that includes a channel director for caching, a disk director for data copying, and an interface for direct communication between computation nodes and storage components, allowing for efficient data access and reducing network latency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If IO operations are performed over a network between host device and storage platform, then data access is enabled, but latency increases
Solution Approach 1:
The patent merges computation resources with storage resources into a unified storage platform. Computation nodes are integrated directly with the storage array, allowing data to be cached and processed within the storage system itself rather than requiring separate host devices to access remote storage over the network. This integration eliminates network hops and reduces IO latency while maintaining data access capability.
2Quantity of substance
If data is stored in slower storage tiers, then storage capacity is increased, but IO performance deteriorates
Solution Approach 1:
The patent implements a caching mechanism within the storage platform where frequently accessed data is pre-loaded into faster cache memory. The channel director intercepts IO requests and serves data from the cache when available, avoiding the need to access slower storage tiers for every IO operation. This preliminary action of caching hot data maintains high IO speeds while preserving the capacity benefits of slower storage tiers.
Solution Approach 2:
The patent creates different performance zones within the storage platform by implementing a hierarchical storage architecture with fast cache memory and slower capacity storage. The channel director intelligently directs IO requests to the appropriate zone based on data location and access patterns, providing local high-speed access for cached data while maintaining access to slower storage tiers for capacity expansion.
3Loss of time
If computation resources are integrated with storage array, then network latency is reduced, but device complexity increases
Solution Approach 1:
The patent designs the storage platform to perform multiple functions: it provides storage capacity, caching, computation, and data processing capabilities within a single integrated system. The computation nodes can execute applications directly on the storage platform, eliminating the need for separate host devices in many scenarios. This multi-functionality consolidates what would otherwise be separate complex systems into one unified platform.
Solution Approach 2:
The storage platform is designed to be self-sufficient by integrating computation resources that can process data and run applications directly within the storage system. The platform can serve its own computation needs and those of connected devices without requiring external host devices, reducing the overall system complexity despite the integration of additional components.
Data Source
AI summary
A storage cloud enabled platform includes storage nodes and application nodes which are the part of the same operating environment. The storage nodes and computation nodes are interconnected inside platform via an interface to provide improved response time and high bandwidth paths for storage applications. The computation nodes may include virtual or non-virtual machines. The interface may support various protocols depending on desired ease of use and response time. For example, a standardized protocol such as SCSI could be used for ease of use and data safety, or a direct data placement protocol could be used to help maximize response rate.