RFL-empirical-study AE

Environment setup

Considering the complexity of the environment setup, we provide a docker container, which can be reached by ssh, to run the code. This repo is already cloned in '/workspace/rfl_empirical_tools'.

# ssh to our docekr environemnt. The details is listed in the `Artifact Location`  section of the ATC AE system.

docker attach rfl_ae

Introduction

We present the data and the code of the RFL empirical study in each directory according to the RQ.

Here is an example of the directory structure.

---- rqN_figure/tableM  
    |---- data  
        rfl_generated_data  
    |---- code  
        preprocess/calculate/plot scripts  
        previous experiment logs(usually stored in the readme.md)
    |---- imgs/table  
        figureM.pdf/tableM_data  
    |---- README.md  
        give instuctions about how to run the code and indicate which figures/table/text this repo corresponds to  
    |---- bash.sh/draw.sh/print.sh [The name does not matter, it is a collection of script to run/plot the data]
        the bash is organized as follows:  
        1) cd /path/to/rfl-path/rq1  
        2) call scripts  
        3) plot.py  
        4) cp data/imgs  

In each directory, you can follow the README.md to reproduce the results. All the code we introduce in the paper is in the code directory. To simplify the reproduction process, we provide a remote docker container.

special cases

Most of the RQs can be reproduced in the docker container. However, some measurement experiments may not be able to run in the docker container because they need specific hardware, such as the NVME disk and e1000 NIC. We will provide a PC equipped with them. The specific steps to reproduce the results will be provided in the README.md in the corresponding directory.

In these experiments, it may need to reboot the PC to change the kernel version to support the driver. If you don't have nvme/e1000 device, you can send an issue to let reboot and select right kernel. We will provide the detailed steps in the README.md in the corresponding directory.