Hybrid Cloud and HPC Approach to High-Performance Dataframes

TL;DR

This paper presents a hybrid cloud and HPC approach to enhance distributed dataframes, integrating UCX to improve compatibility and performance in diverse environments beyond traditional MPI-based systems.

Contribution

We integrated UCX as a communication layer in Cylon to address MPI compatibility issues, enabling better deployment in varied HPC and cloud environments.

Findings

UCX integration improves compatibility with non-MPI environments

Enhanced performance in distributed data processing tasks

Method applicable to other MPI-dependent applications

Abstract

Data pre-processing is a fundamental component in any data-driven application. With the increasing complexity of data processing operations and volume of data, Cylon, a distributed dataframe system, is developed to facilitate data processing both as a standalone application and as a library, especially for Python applications. While Cylon shows promising performance results, we experienced difficulties trying to integrate with frameworks incompatible with the traditional Message Passing Interface (MPI). While MPI implementations encompass scalable and efficient communication routines, their process launching mechanisms work well with mainstream HPC systems but are incompatible with some environments that adopt their own resource management systems. In this work, we alleviated this issue by directly integrating the Unified Communication X (UCX) framework, which supports a variety of classic HPC and non-HPC process-bootstrapping mechanisms as our communication framework. While we experimented with our methodology on Cylon, the same technique can be used to bring MPI communication to other applications that do not employ MPI's built-in process management approach.