Thursday, November 9, 2017

Data on SNET and a few suggestions for ETECSA

What would be the impact of, say, a $100,000 equipment grant from ETECSA to SNET?

I've written several posts on Cuba's user-deployed street networks, the largest of which is SNET in Havana. (SNET was originally built by the gaming community, but the range of services has grown substantially). My posts and journalist's accounts like this one describe SNET, but a new paper presents SNET measurement data as well as descriptive material.

The abstract of the paper sums it up:
Working in collaboration with SNET operators, we describe the network’s infrastructure and map its topology, and we measure bandwidth, available services, usage patterns, and user demographics. Qualitatively, we attempt to answer why the SNET exists and what benefits it has afforded its users. We go on to discuss technical challenges the network faces, including scalability, security, and organizational issues.
You should read the paper -- it's interesting and well-written -- but I can summarize a few points that caught my attention.

SNET is a decentralized network comprised of local nodes, each serving up to 200 users in a neighborhood. The users connect to local nodes using Ethernet cables strung over rooftops, etc. or WiFi. The local nodes connect to regional "pillars" and the pillars peer with each other over fixed wireless links. The node and pillar administrators form a decentralized organization, setting policy, supporting users and keeping their servers running and online as best they can. (This reminds me of my school's first Web server -- a Windows 3 PC on my desk that crashed frequently).

SNET organization (source)

The average utilized bandwidth between two pillars during a 24-hour period was 120 Mb/s of a maximum throughput of 250 Mb/s and the authors concluded that throughput is generally constrained by the available bandwidth in the WiFi links between pillars. As such, faster inter-pillar links and/or adding new pillars would improve performance. Faster links from local nodes to pillars, new node servers, etc. would also add to capacity and availability, but that hardware would cost money. The Cuban government would probably see the provision of outside funds as subversive, but what would be the impact of, say, a $100,000 equipment grant from ETECSA to SNET?

The paper drills down on the network topology, discusses applications and presents usage and performance statistics. Forums are one of the applications and one of the forums is Netlab, a technical community of over 6,000 registered members who have made over 81,000 posts. They focus on open-source development and have written a SNET search engine and technical guides on topics like Android device repair. The export of Cuban content and technology has been a long-standing focus of this blog, and it would be cool to see Netlab available to others on the open Internet.

Netlab forum growth

The authors of the paper say that as far as they know, "SNET is the largest isolated community-driven network in existence" (my italics). While it may be the largest isolated community network there are larger Internet-connected community networks and that is a shame. I hope Cuba plans to "leapfrog" to next-generation technology and policy) while implementing stopgap measures like WiFi hotspots, 3G mobile and DSL. If SNET and other community networks were legitimized, supported and linked to the Internet (or even the Cuban intranet), they would be useful stopgap technology. ETECSA could also use the skills of the street net builders.

I don't expect ETECSA to take my advice, but if working with SNET is too big a step, they might test community collaboration by working with the developers of a smaller street net like the one in Gaspar or try involving communities in networking some schools, experimenting with community-installed backhaul or deploying interim satellite connectivity.

(You can find links to the paper, Initial Measurements of the Cuban Street Network, presentation slides and abstract here).
Real Time Analytics