maxi-k / costoptimal-model

Model implementation and explorative UI for the paper "Towards Cost-Optimal Query Processing in the Cloud". Slides: https://bit.ly/37ZfePj

Towards Cost-Optimal Query Processing in the Cloud - Model Implementation

The implementation for the cost model introduced in the paper "Towards Cost-Optimal Query Processing in the Cloud".

Running

This version of the model is implemented in R. As such you need to have R installed on your computer, see here. We also depend on various R packages that are not part of the basic installation. These should install automatically when running any file. If they don't, please look at the list in util.r#util.packages.basic.setup and install them manually.

To start an interactive, browser-based explorer for the model predictions, please execute the file app.r by following these steps:

1. Run R in a command line. This should open an R REPL.
2. Type source("app.r") in the R REPL. This will load the source file and should open your browser with the URL http://0.0.0.0:3030, displaying the model explorer.

To generate the plots used in the paper, please execute the respective code blocks in the plots.r file.

Project Structure

Files of interest:

Path	Description
aws.r	Code for loading and manipulating aws instance data from https://ec2instances.info
model.r	The main model implementation as described in the paper
elasticity.r	The greedy auto-scaling algorithm and accompanying plots
app.r	The interactive web-based UI for exploring model predictions
plots.r	ggplot sources for the plots shown in the paper

The main algorithms relating to the model section in the paper are implemented in the functions model.calc.costs and model.calc.time.for.config in model.r.

Note that this is exploratory code for research purposes; code quality and maintainability are secondary.