基于WebAssembly的FaaS框架，具备分布式机器学习能力 | KC23

WebAssembly-Based FaaS Framework with Distributed Machine Learning... - Wilson Wang & Michael Yuan

基于WebAssembly的FaaS框架，具备分布式机器学习能力 | WebAssembly-Based FaaS Framework with Distributed Machine Learning Capabilities - Wilson Wang, ByteDance & Michael Yuan, Second State

我们的目标是创建一个利用WebAssembly的FaaS平台，这是一种安全且轻量级的技术，用于机器学习任务，特别是推理作业。为了实现这一目标，我们正在将WebAssembly与Ray集成，Ray是一个广泛使用的用于扩展人工智能和Python应用程序的框架，以创建一个具有分布式机器学习能力的强大FaaS平台。Ray的高级功能，如分布式调度、对象存储和任务间通信，使其成为一个优秀的ML-enabled FaaS平台选择，它统一了资源抽象，并消除了不同FaaS函数之间的障碍。通过将WebAssembly与Ray集成，我们可以使任务更加轻量级，并扩展Ray对编程语言（如Rust、Go和JavaScript）的支持，以简化现有应用程序的移植过程。

Our goal is to create a FaaS platform that utilizes WebAssembly, a secure and lightweight technology, for Machine Learning tasks, specifically inference jobs. To achieve this, we are integrating WebAssembly with Ray, a widely-used framework for scaling AI and Python applications, to create a powerful FaaS platform with distributed Machine Learning capabilities. Ray's advanced features, such as distributed scheduling, object storage, and inter-task communication, make it an excellent choice for an ML-enabled FaaS platform that unifies resource abstraction and eliminates barriers between different FaaS functions. By integrating WebAssembly with Ray, we can make tasks even more lightweight, and expand Ray's support for programming languages, such as Rust, Go, and JavaScript, to simplify the process of porting existing applications.

CNCF概况（幻灯片）

扫描二维码联系我们！

CNCF (Cloud Native Computing Foundation)成立于2015年12月，隶属于Linux  Foundation，是非营利性组织。 

CNCF（云原生计算基金会）致力于培育和维护一个厂商中立的开源生态系统，来推广云原生技术。我们通过将最前沿的模式民主化，让这些创新为大众所用。请关注CNCF微信公众号。