This repository contains the aggregate data and the scripts needed to re-generate every plot in the accepted IMC'22 paper, titled "Respect the ORIGIN! A Best-case Evaluation of Connection Coalescing in the Wild".
By: Sudheesh Singanamalla, Muhammad Talha Paracha, Suleman Ahmad Jonathan Hoyland, Luke Valenta, Yevgen Safronov, Peter Wu, Andrew Galloni, Kurtis Heimerl, Nick Sullivan, Christopher A. Wood, Marwan Fayed
$ python generate.pyThe raw dataframes needed for each plot is available in data and the generated figures
are stored in the plots directory.
| # | Filename | Columns | Description |
|---|---|---|---|
| Figure 1 | figure-1-dataframe-asn-share.csv | ASes | Number of Unique ASes Contacted |
| Count | Percentage of websites contacting the number of ASes | ||
| figure-1-dataframe-cdf.csv | cdf-X | Number of unique ASes Contacted | |
| cdf-Y | Cumulative Sum of percentage of unique ASes Contacted | ||
| Figure 2 | figure-2.drawio | Pre-rendered informational image of timeline reconstruction | |
| Figure 3 | figure-3-dataframe.csv | dns | Number of DNS requests made during page load |
| tls | Number of TLS requests made during page load | ||
| ip_dns_tls | Modelled expectation of DNS and TLS request counts in IP based coalescing | ||
| origin_dns_tls | Modelled expectation of DNS and TLS request counts in ORIGIN Frame based coalescing | ||
| Figure 4 | figure-4-dataframe.csv | original_san_list | Number of DNS SAN Entries from certificates in measurement |
| updated_san_list | Number of DNS SAN Entries from modelled ideal certificates | ||
| Figure 5 | figure-5-dataframe.csv | original_san_list | Number of DNS SAN Entries from certificates in measurement |
| updated_san_list | Number of DNS SAN Entries from modelled ideal certificates | ||
| mapped_san_additions | Number of additions to make to original_san_list, each row is a certificate mapping | ||
| Figure 6 | figure-6.drawio | Pre-rendered informational image about certificate changes during experiments | |
| Figure 7a | live_ip_coalescing_experiment_control.csv | CDNJS-Coalesced | Number of requests successfully coalescing cdnjs sub resources |
| Figure 7b | live_origin_coalescing_experiment_control.csv | CDNJS-Not-Coalesced | Number of requests failed to coalesce cdnjs sub resources |
| Both the files have the same schema. | Total-TLS | Number of total TLS requests made | |
| Type | Categorical variable - Experiment | Control type | ||
| Figure 8 | origin_frame_control_timeline.csv | Time | Date format YYYY-MM-DD |
| origin_frame_experiment_timeline.csv | Count | Number of TLS connections per second (100k) | |
| Figure 9a | figure-9a-dataframe.csv | page_load_times | Measured page load time (ms) |
| ip_page_load_times | Modelled IP based coalescing page load time (ms) | ||
| as_page_load_times | Modelled ORIGIN based coalescing page load time (ms) | ||
| cf_only_page_load_times | Modelled ORIGIN based coalescing if only Cloudflare deploys changes (ms) | ||
| Figure 9b | figure-9b-dataframe.csv | Mode | Categorical variable - Experiment | Control type |
| Time | Measured page load time in active ORIGIN Frame experiments. |