Cirrus

Cirrus is a serverless machine learning library. Cirrus provides a list of machine learning algorithms that can scale to many serverless lambdas in the cloud.

Requirements

The Cirrus backend has been tested on Ubuntu 14.04/16.04/18.04 and Amazon AMI.

It has been tested with the following environment / dependencies:

• g++-7

In the Amazon AMI please do:

$ sudo yum install glibc-static
$ sudo yum install openssl-static.x86_64
$ sudo yum install zlib-static.x86_64

In Ubuntu please do:

$ sudo apt-get install build-essential cmake automake zlib1g-dev libssl-dev libcurl4-nss-dev bison libldap2-dev libkrb5-dev

Building

$ ./bootstrap.sh
$ make -j 10

Paper

This project is part of a research project on Serverless Machine Learning Workflows. This works has been published and can be found here:

Joao Carreira, Pedro Fonseca, Alexey Tumanov, Andrew Zhang, Randy Katz. In the ACM Symposium on Cloud Computing 2019 (SoCC'19)

Funding

This work has been generously supported by AWS Cloud Research, FCT (Portuguese Science Foundation), NSF CISE Expeditions Award CCF-1730628, and gifts from Alibaba, Amazon Web Services, Ant Financial, CapitalOne, Ericsson, Facebook, Futurewei, Google, Intel, Microsoft, Nvidia, Scotiabank, Splunk and VMware.

Contributors

Joao Carreira, Andrew Zhang, Jeff Yu, Ryan Wang, Neel Somani, Shea Conlon, Andy Wang, Pedro Fonseca, Alexey Tumanov, Randy Katz