Syria Internet Infrastructure Study
This project reads what public platforms already recorded about Syria’s internet: routing, outages and censorship. It has never sent a packet to a Syrian network. Every observation was collected by someone else’s platform, for its own purposes.
The collectors
Python collector scripts, one per kind of record. The main sources:
| Source | What it holds |
|---|---|
| RIPE RIS, RouteViews | BGP announcements and withdrawals seen by route collectors |
| RIPEstat | The same RIS data through an API, plus registry data |
| IODA | Outage signals per country and network |
| OONI | Censorship measurements run by volunteers |
| Cloudflare Radar | Traffic trends and outage annotations |
The collectors log the queries they send. The early ones logged a query before sending it and not the result, so part of that log says not_recorded, and it’s published with those gaps.
The withdrawals that weren’t there
RIPEstat’s bgp-updates endpoint can be queried for a whole AS. Run that way, it returns announcements and no withdrawals. A withdrawal carries no AS path, so there’s nothing to match the AS against.
I wrote to the RIPE NCC about it. Their reply, in September 2026: “you are correct that the lack of path info in a withdrawal makes it nontrivial to return them when querying for an AS”, and they would adjust the documentation. The collectors now query per prefix, which returns both.
In the same exchange I asked for permission to redistribute RIS data in the dataset, and they granted it, with attribution to the RIPE NCC.
What’s public
The dataset is on Zenodo, published 2026-09-21. Everything in it derives from RIPE RIS: my files are CC BY 4.0, and the RIPE data stays under the RIPE NCC’s terms. The code repository is private for now.