Message-Based Code Execution via Key-Value Storage
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional computing system architectures face inefficiencies due to latency and throughput limitations introduced by data communication buses, particularly when storage resource throughput exceeds data bus capabilities, leading to delays in accessing data for computing resources.
Innovation Solution
A message-based architecture utilizing a key-value storage system where computing resources access and merge executable code with stored data, leveraging NVMe protocol and PCIe transport to minimize latency by integrating compute and storage resources within computing nodes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If traditional data bus architecture is used to connect computing resources and storage resources, then system simplicity is maintained, but latency increases and throughput is limited when storage throughput exceeds data bus capabilities
Solution Approach 1:
The system is segmented into multiple computing nodes, each with integrated storage resources. This segmentation allows each node to access its local storage directly without competing for shared data bus resources, thereby reducing latency while maintaining overall system simplicity through modular design.
Solution Approach 2:
The patent transitions from a centralized data bus architecture to a distributed architecture where computing and storage resources are coupled at the node level. This dimensional change from centralized to distributed connectivity provides alternative access paths and eliminates the data bus bottleneck without requiring complete architectural overhaul.
2Productivity
If storage resources are designed with high throughput capabilities, then data availability is improved, but data access latency increases due to data bus limitations
Solution Approach 1:
Computing resources and storage resources are merged within the same computing node, creating a tight coupling that eliminates data bus intermediate steps. This merging allows high-throughput storage resources to be directly accessed by computing resources at minimal latency, as they share the same node infrastructure and memory space.
3Ease of operation
If computing resources frequently access stored data, then computational tasks can be executed, but efficiency decreases due to data bus latency and throughput limitations
Solution Approach 1:
Data is pre-loaded into local storage resources within the computing node before computational tasks require it. This preliminary action ensures that when computing resources need to access data, it is already available in high-speed local storage rather than requiring retrieval through the slower data bus, thereby maintaining ease of access while improving computational efficiency.
Data Source
AI summary
Methods and systems for executing code are provided. A message with executable code is received by a processor that utilizes the message to generate a key associated with a value having executable code stored at a key-value store. The processor retrieves the value from the key-value store using the generated key. The processor then merges the executable code in the received message and the executable code in the retrieved value to generate a merged value for code execution.