A novel approach pyStorageLess introduces storage interoperability for serverless functions in federated serverless infrastructures. The serverless functions are deployed only once, and the storage can be dynamically linked to the functions at runtime by the user through control data inputs while invoking the serverless functions. pyStorageLess uses the Von Neumann approach to abstract the function to have computing resources, memory, storage, and input/output data, regardless of the provider that hosts each part. pyStorageLess splits data inputs into value and control, allowing users to dynamically attach the storage at runtime. With such extreme flexibility, users may distribute parts of the function across different providers to improve performance.

错误:搜索内容不能为空,请输入英文关键词
错误:关键词超出字数限制,请精简
高级检索

pyStorageLess: Leveraging Von Neumann’s Architecture to Abstract Storage Heterogeneity in Serverless Applications

  • Sashko Ristov,
  • Mika Hautz,
  • Philipp Gritsch,
  • Isabella Schmut,
  • Peter Koll,
  • Michael Felderer

摘要

A novel approach pyStorageLess introduces storage interoperability for serverless functions in federated serverless infrastructures. The serverless functions are deployed only once, and the storage can be dynamically linked to the functions at runtime by the user through control data inputs while invoking the serverless functions. pyStorageLess uses the Von Neumann approach to abstract the function to have computing resources, memory, storage, and input/output data, regardless of the provider that hosts each part. pyStorageLess splits data inputs into value and control, allowing users to dynamically attach the storage at runtime. With such extreme flexibility, users may distribute parts of the function across different providers to improve performance.