This repository contains all relevant artifacts required to reproduce the experiments conducted for the paper work "Prebaking Functions to Warm the Serverless Cold Start". This repository also holds the data, and the data analysis scripts used to generate the paper main results.
The module serverless-handlers contains the required software to conduct the experiments described in the paper. The README.md outlines how to run an experiment to measure serverless applications start-up time.
However, in this section we describe the required steps to reproduce all the paper experiments.
First of all, you need to clone this repository by running the following commands:
git clone --recurse-submodules https://github.com/paulofelipefeitosa/prebaking-middleware-2020
cd prebaking-middleware-2020
After it, you need to install the following linux tools:
libcurl
,g++
,wget
;R >= 4.0.0
.
Also, it is required to install the following R packages:
dplyr
,ggplot2
,reshape
andggpubr
.
Finally, you need to install all serverless-handlers
dependencies that
were describe in the section
dependencies.
Please, note that we performed our experiments using Ubuntu 16.04
with Linux kernel
4.15.0-45-generic-x86_64
. However, we already reproduced the experiments using
Debian 10 with Linux kernel 4.19.0-12
.
The Function Start-up Time
experiments reference to all executed experiments to generate
the results for the paper section 4.2
.
Here you can access all required steps to reproduce
the Function Start-up Time
experiments.
The Start-up Components
experiments reference to all conducted experiments to
generate the results for the paper section 4.2.1
.
Here you can access all required steps to reproduce
the Start-up Components
experiments.
The Choosing Ingredients
experiments reference to all conducted experiments to
generate the results for the paper section 4.2.2
.
Here you can access all required steps to reproduce
the Choosing Ingredients
experiments.
The data along with the analysis scripts are stored inside the eval
directory. There we explain how to analyze the data
and generate the paper main results.