Networked Storage Architecture Dynamic Workload Allocation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In storage architectures where server computers are directly connected to storage devices, processing bottlenecks can occur due to high data processing activity exceeding available resources, making it difficult to increase processing capability or storage capacity without physical hardware changes.
Innovation Solution
Implementing a system that dynamically associates server computers with network-enabled storage devices via a computer network, using a workload manager to allocate servers and storage resources based on changing workload demands, allowing for flexible and scalable storage by transmitting data block-level access commands over the network.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If server computers are directly connected to storage devices, then data access speed and processing capability are improved, but processing bottlenecks occur when data processing activity exceeds available resources
Solution Approach 1:
The patent introduces a network as an intermediary layer between server computers and storage devices. Instead of direct physical connections, servers communicate with storage devices through network protocols (iSCSI, Fibre Channel over Ethernet). This intermediary network infrastructure distributes processing loads and prevents single-point bottlenecks, allowing multiple servers to access multiple storage devices simultaneously without overwhelming any single connection.
Solution Approach 2:
The patent creates a universal network-based storage access system where any server can access any storage device through standard network protocols. This multi-functional architecture allows the same infrastructure to handle diverse workloads (data processing, backup, replication, analytics) without requiring dedicated physical connections for each function, thereby improving resource utilization and system reliability under varying loads.
2Productivity
If additional processing capability or storage capacity is needed, then system performance can be improved, but physical hardware changes are required which reduce flexibility
Solution Approach 1:
The patent replaces mechanical/physical hardware connections with software-based network communication. Instead of physically connecting servers to storage devices through cables and interfaces, the system uses network protocols and software drivers to establish logical connections. This substitution allows dynamic allocation and reconfiguration of resources through software without any physical hardware changes, greatly enhancing system flexibility and adaptability.
Solution Approach 2:
The patent implements dynamic resource allocation where servers and storage devices can be associated and disassociated through software configuration rather than physical reconnection. The system can dynamically adjust which servers access which storage devices based on workload demands, enabling flexible scaling of processing capability and storage capacity without permanent hardware commitments.
3Device complexity
If all data access operations are channeled through a single server computer, then simplified management is achieved, but processing bottlenecks occur due to resource limitations
Solution Approach 1:
The patent segments the centralized data access model into a distributed architecture where multiple servers can simultaneously access multiple storage devices through the network. Instead of one server handling all access operations, the system divides workloads across multiple server-storage pairs, with each server managing its own access operations independently. This segmentation eliminates the single-point bottleneck while maintaining manageable complexity through standardized network protocols.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Technologies are provided for a dynamically scalable storage architecture. A workload manager can be configured to monitor compute workloads targeting network-enabled storage devices. The workload manager can identify servers to perform all or part of the compute workloads. The workload management server can configure the identified servers to establish logical connections with the network-enabled storage devices, wherein the identified servers can transmit data access commands to the network-enabled storage devices via a computer network. The identified servers can then be used to perform all or part of the compute workloads targeting the network-enabled storage devices. In at least some embodiments, the workload manager can monitor a pool of available servers, from which servers can be identified and associated with network-enabled storage devices. In a different or further embodiment, the workload management server can instantiate new virtual servers and associate the virtual servers with network-enabled storage devices.