Over the years, Alluxio has grown significantly to be the data orchestration framework for the cloud. The community developers and users have contributed a lot of effort and innovation to make Alluxio the system it is today. There are many users and companies deploying Alluxio at very large scale, and with the large scale, comes different types of challenges.
In this talk, I will introduce the high-level architecture of the current system, and present the various components of Alluxio. Also, I will discuss some of the main challenges of large scale Alluxio deployments, and the lessons we learned from those environments. This talk will detail some of the major scalability improvements added in the past several months, and how users can benefit from the changes.
Over the years, Alluxio has grown significantly to be the data orchestration framework for the cloud. The community developers and users have contributed a lot of effort and innovation to make Alluxio the system it is today. There are many users and companies deploying Alluxio at very large scale, and with the large scale, comes different types of challenges.
In this talk, I will introduce the high-level architecture of the current system, and present the various components of Alluxio. Also, I will discuss some of the main challenges of large scale Alluxio deployments, and the lessons we learned from those environments. This talk will detail some of the major scalability improvements added in the past several months, and how users can benefit from the changes.
Video:
Presentation Slides:
Complete the form below to access the full overview:
.png)
Videos
TorchTitan is a proof-of-concept for Large-scale LLM training using native PyTorch. It is a repo that showcases PyTorch's latest distributed training features in a clean, minimal codebase.
In this talk, Tianyu will share TorchTitan’s design and optimizations for the Llama 3.1 family of LLMs, spanning 8 billion to 405 billion parameters, and showcase its performance, composability, and scalability.
As large-scale machine learning becomes increasingly GPU-centric, modern high-performance hardware like NVMe storage and RDMA networks (InfiniBand or specialized NICs) are becoming more widespread. To fully leverage these resources, it’s crucial to build a balanced architecture that avoids GPU underutilization. In this talk, we will explore various strategies to address this challenge by effectively utilizing these advanced hardware components. Specifically, we will present experimental results from building a Kubernetes-native distributed caching layer, utilizing NVMe storage and high-speed RDMA networks to optimize data access for PyTorch training.