Skip to content

FaaSNet: Scalable and Fast Provisioning of Custom Serverless Container Runtimes at Alibaba Cloud Function Compute (USENIX ATC'21)

License

Notifications You must be signed in to change notification settings

ds2-lab/FaaSNet

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

26 Commits
 
 
 
 
 
 
 
 
 
 

Repository files navigation

FaaSNet

FaaSNet is the first system that provides an end-to-end, integrated solution for FaaS-optimized container runtime provisioning. FaaSNet uses lightweight, decentralized, and adaptive Function Trees (FTs) to avoid major platform bottlenecks.

Our USENIX ATC'21 paper: FaaSNet: Scalable and Fast Provisioning of Custom Serverless Container Runtimes at Alibaba Cloud Function Compute

Download the preprint version on arXiv.

This repo contains two components:

The Function Cold Start Traces from Alibaba Cloud Function Compute

Introduction

Our trace dataset is a subset of the data described in, and analyzed, in our ATC '21 paper. The traces were obtained by collecting 24-hour production-level logs from two datacenters (Beijing and Shanghai) in Alibaba Function Compute service during May 2021.

Using the Data

License

The data is made available and licensed under an Apache License 2.0. By downloading it or using them, you agree to the terms of this license.

Dataset Downloading

We would like you could help us take the survey before downloading the FaaS cold start traces. The link is as following Google Form, you could retrieve the download link after submmting the form successfully.

Please free to let us know if you have any further questions.

Schema and Description

Field Description
__time__ TimeStamp in seconds
functionName Unique id for the function name
latency Function cold start latency1 in seconds
runtime Function runtime (Python, nodejs, custom-runtime, etc)
memoryMB Function's allocated memory in MB

Notes:

  1. The function cold start latency only counts the system level's latency, such as container initialization, etc, instead of end-to-end cold start latency.

Function Tree Prototype

Our released function tree (FT) prototype is the version that we evaluated in the ATC '21 paper. We are continuing to improve the performance of it. We're happy to accept contributions! Please feel free to hack on the FT and integrate it into your framework/platform :-).

Attribution

If you use our trace dataset and/or the FT prototype for a publication or project, please cite the accompanying paper using this bibtex:

Ao Wang, Shuai Chang, Huangshi Tian, Hongqi Wang, Haoran Yang, Huiba Li, Rui Du, Yue Cheng. "FaaSNet: Scalable and Fast Provisioning of Custom Serverless ContainerRuntimes at Alibaba Cloud Function Compute", in Proceedings of the 2021 USENIX Annual Technical Conference (USENIX ATC 21). USENIX Association, July 2021.

Lastly, if you have any questions, comments, or concerns, or if you would like to share tools for working with the traces, please contact us at [email protected].

To Cite FaaSNet

@inproceedings {273798,
author = {Ao Wang and Shuai Chang and Huangshi Tian and Hongqi Wang and Haoran Yang and Huiba Li and Rui Du and Yue Cheng},
title = {FaaSNet: Scalable and Fast Provisioning of Custom Serverless Container Runtimes at Alibaba Cloud Function Compute},
booktitle = {2021 {USENIX} Annual Technical Conference ({USENIX} {ATC} 21)},
year = {2021},
isbn = {978-1-939133-23-6},
pages = {443--457},
url = {https://www.usenix.org/conference/atc21/presentation/wang-ao},
publisher = {{USENIX} Association},
month = jul,
}

About

FaaSNet: Scalable and Fast Provisioning of Custom Serverless Container Runtimes at Alibaba Cloud Function Compute (USENIX ATC'21)

Topics

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages