A data processing optimization method, device, medium and program product

By creating a distributed shared storage system in the HPC cluster, the problems of low data interaction efficiency and high storage cost between computing nodes and visualization nodes are solved, achieving cross-system compatibility and low-cost data sharing.

CN122363601APending Publication Date: 2026-07-10SHANGHAI SILANG WANWEI COMPUTING TECH CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
SHANGHAI SILANG WANWEI COMPUTING TECH CO LTD
Filing Date
2025-07-30
Publication Date
2026-07-10

AI Technical Summary

Technical Problem

The existing HPC cluster suffers from inefficient data interaction between compute nodes and visualization nodes, poor cross-system compatibility, and high storage costs, resulting in long data migration times and increased storage costs.

Method used

Create a distributed shared storage system for HPC clusters, including a first cold data pool, a second cold data pool, and a hot data pool. Process data through compute nodes and visualization nodes, and achieve cross-system compatibility of data under different operating systems.

Benefits of technology

It improves data interaction efficiency, reduces storage costs, enables cross-system data sharing and multi-system compatibility, and reduces data migration time.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122363601A_ABST
    Figure CN122363601A_ABST
Patent Text Reader

Abstract

This invention discloses a data processing optimization method, device, medium, and program product, relating to the field of computer storage. The data processing optimization method includes: creating a distributed shared storage system for a high-performance computing (HPC) cluster; storing raw cold data in a first cold data pool, and retrieving data to be computed from the first cold data pool based on current data computation instructions; storing preliminary result data obtained by processing the data to be computed through computing nodes in a hot data pool; obtaining target result data by processing the preliminary result data through visualization nodes, and storing the target result data in a second cold data pool; wherein the computing nodes and visualization nodes are cross-system nodes in the HPC cluster. The technical solution of this invention can complete data interaction between computing nodes and visualization nodes based on a distributed shared storage system, possessing multi-system compatibility and low storage cost.
Need to check novelty before this filing date? Find Prior Art