Fabric-Attached Memory for Ephemeral Data in Cloud Systems
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Cloud computing systems face increased data processing complexity and processing errors due to the large number of read and write data operations performed on remote storage devices for ephemeral intermediate data produced by daisy-chained compute tasks.
Innovation Solution
The system employs cloud orchestrator circuitry to determine task graphs, data graphs, and function popularity heap parameters, and fabric manager circuitry to allocate programmable IC devices and fabric-attached memory (FAM) for storing ephemeral data, thereby reducing the number of data operations with remote storage devices.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If processing units perform many read and write data operations to remote storage devices, then data can be stored and retrieved, but data processing complexity and processing errors increase
Solution Approach 1:
The patent introduces fabric-attached memory as an intermediary layer between processing units and remote storage devices. This memory fabric serves as a local buffer that holds ephemeral intermediate data, reducing the frequency of read/write operations to remote storage and thereby decreasing data processing complexity and errors while maintaining data integrity.
2Reliability
If processing units perform many read and write data operations to remote storage devices, then data can be stored and retrieved, but power consumption increases
Solution Approach 1:
The fabric-attached memory acts as an intermediary that retains ephemeral data locally during processing operations. By keeping intermediate data in this local memory fabric rather than repeatedly accessing remote storage, the system maintains data persistence requirements while significantly reducing power consumption associated with frequent I/O operations.
3Reliability
If processing units perform many read and write data operations to remote storage devices, then data can be stored and retrieved, but processing efficiency decreases
Solution Approach 1:
The system performs preliminary actions by pre-loading ephemeral intermediate data into the fabric-attached memory before processing operations begin. This preliminary placement of data in accessible memory allows processing units to efficiently access and manipulate data without repeated remote storage I/O operations, thereby improving processing efficiency while maintaining data integrity through the fabric-attached memory's reliable storage capabilities.
Data Source
AI summary
A cloud computing system includes cloud orchestrator circuitry and fabric manager circuitry. The cloud orchestrator circuitry receives an input application and determines a task graph, a data graph, and a function popularity heap parameter for the input application. The task graph comprises an indication of function interdependency of functions of the input application, the data graph comprises an indication of data interdependency of the functions, and the function popularity heap parameter corresponds to a re-usability index for the functions. The fabric manager circuitry allocate a first programmable integrated circuit (IC) device to perform a first function of the input application based on the task graph, the data graph, and the function popularity heap parameter.