GPEC: An Open Emulation Platform to Evaluate GPU/ML Workloads on Erasure Coding Storage

Thu, 08 Feb 2024 00:00:00 +0000

Project Idea Description

Topics: Storage Systems, Machine Learning, Erasure Coding
Skills: C/C++, Python, PyTorch, Bash scripting, Linux, Erasure Coding, Machine Learning
Difficulty: Hard
Size: Large (350 hours)
Mentors: Meng Wang (primary contact), John Bent

Large-scale data centers store immense amounts of user data across a multitude of disks, necessitating redundancy strategies like erasure coding (EC) to safeguard against disk failures. Numerous research efforts have sought to assess the performance and durability of various erasure coding approaches, including single-level erasure coding, locally recoverable coding, and multi-level erasure coding.

Despite its widespread adoption, a significant research gap exists regarding the performance of large-scale erasure-coded storage systems when exposed to machine learning (ML) workloads. While conventional practice often leans towards replication for enhanced performance, this project seeks to explore whether cost-effective erasure encoding can deliver comparable performance. In this context, several fundamental questions remain unanswered, including: Can a typical erasure-coded storage system deliver sufficient throughput for ML training tasks? Can an erasure-coded storage system maintain low-latency performance for ML training and inference workloads? How does disk failure and subsequent repair impact the throughput and latency of ML workloads? What influence do various erasure coding design choices, such as chunk placement strategies and repair methods, have on the aforementioned performance metrics?

To address these questions, the most straightforward approach would involve running ML workloads on large-scale erasure coded storage systems within HPC data centers. However, this presents challenges for researchers and students due to limited access to expensive GPUs and distributed storage systems, especially when dealing with large-scale evaluations. Consequently, there is a need for a cost-effective evaluation platform.

The objective of this project is to develop an open-source platform that facilitates cheap and reproducible evaluations of erasure-coded storage systems concerning ML workloads. This platform consists of two key components: GPU Emulator: This emulator is designed to simulate GPU performance for ML workloads. Development of the GPU emulator is near completion. EC Emulator: This emulator is designed to simulate the performance characteristics of erasure-coded storage systems. It is still in the exploratory phase and requires further development.

The student’s responsibilities will include documenting the GPU emulator, progressing the development of the EC emulator, and packaging the experiments to ensure easy reproducibility. It is anticipated that this platform will empower researchers and students to conduct cost-effective and reproducible evaluations of large-scale erasure-coded storage systems in the context of ML workloads.

Project Deliverable

Build an EC emulator to emulate the performance characteristics of large-scale erasure-coded storage systems
Incorporate the EC emulator into ML workloads and GPU emulator
Conduct reproducible experiments to evaluate the performance of erasure-coded storage systems in the context of ML workloads
Publish a Trovi artifact shared on Chameleon Cloud and a GitHub repository with open-source code

Reproducible Evaluation of Multi-level Erasure Coding

Thu, 02 Feb 2023 00:00:00 +0000

Massive storage systems rely heavily on erasure coding (EC) to protect data from drive failures and provide data durability. Existing storage systems mostly adopt single-level erasure coding (SLEC) to protect data, either performing EC at the network level or performing EC at the local level. However, both SLEC approaches have limitations, as network-only SLEC introduces heavy network traffic overhead, and local-only SLEC cannot tolerate rack failures.

Accordingly, some data centers are starting to use multi-level erasure coding (MLEC), which is a hybrid approach performing EC at both the network level and the local level. However, prior EC research and evaluations mostly focused on SLEC, and it remains to be answered how MLEC is compared to SLEC in terms of durability, capacity overhead, encoding throughput, network traffic, and other overheads.

Therefore, in this project we seek to build a platform to evaluate the durability and overheads of MLEC. The platform will allow us to evaluate dozens of EC strategies in many dimensions including recovery strategies, chunk placement choices, various parity schemes, etc. To the best of our knowledge, there is no other evaluation platform like what we propose here. We seek to make the platform open-source and the evaluation reproducible, allowing future researchers to benefit from it and conduct more research on MLEC.

Building a platform to evaluate MLEC

Topics: Storage systems, reproducibility, erasure coding, evaluation
Skills: Linux, C, Python
Difficulty: Medium
Size: 350 hours
Mentor(s): John Bent and Anjus George
Contributor(s): Zhiyan "Alex" Wang

Build a platform to evaluate the durability and overheads of MLEC. The platform will be able to evaluate different EC strategies in various dimensions including repair strategies, chunk placement choices, parity schemes, etc. Analyze the evaluation results.

Specific tasks:

Work with mentors on understanding the context of the project.
Reproduce the SLEC evaluation results from prior SLEC evaluation tools
Based on prior SLEC evaluation tools, build a platform to evaluate the durability and overheads of MLEC under various EC strategies
Collect and analyze the MLEC evaluation results

John Bent | UCSC OSPO

GPEC: An Open Emulation Platform to Evaluate GPU/ML Workloads on Erasure Coding Storage

Reproducible Evaluation of Multi-level Erasure Coding

Building a platform to evaluate MLEC