Edge traces

This repository contains a Jupyter notebook that can be used to generate traces for research in the area of edge computing.

The generated traces are based on the Alibaba cluster-trace-microservices-v2021 dataset. These traces are not geo-localized, so the notebook takes the traces for the most used microservices and considers each one as the trace of a different app in a different region.

To generate the trace file, you have to run the notebook:

pip install -r requirements.txt
jupyter notebook

It downloads the original traces, processes them, and generates a new file called edge_1h.csv. This file contains data grouped by hour. This is the start of the file:

timestamp,reg,app,reqs
1970-01-01 00:00:00,0,0,4827602
1970-01-01 01:00:00,0,0,11731750
1970-01-01 02:00:00,0,0,22109809
1970-01-01 03:00:00,0,0,15124156
...

This data can be used with the Edarop package to obtain optimal allocations in edge architectures.