"Due to both privacy and efficiency concerns, almost all Ethernets today are fully switched"
I don't think those were the reasons. The main reason I remember was cost. Shared media pretty much meant coax, and twisted pair is much cheaper. When switches became cheap enough and capable enough, the savings of running TP everywhere became the dominant factor. This also eliminated long cable runs and the frequent disruption of losing an entire segment at once, so reliability was also a factor, but IIRC those improvements were secondary in most purchasers' minds.
I guess cost reduction is a form of efficiency, but I certainly never heard privacy as a reason before.
I think they are referring to switches vs. hubs (repeaters). If you have been on an Ethernet segment that uses hubs instead of switches, you will remember that tcpdump will allow you to see all traffic to all stations. That was typically also in the days of unencrypted email and web traffic
I keep a hub around for just that purpose. The downside is that Gigabit eliminated half duplex for all practical purposes, so you have to use a tap or SPAN on anything faster.
Privacy became a reason when ISPs started providing internet access over Ethernet. As hubs and unmanaged switches are vulnerable to sniffing, ISPs had to either give everyone separate VLANs or use managed switches with protections from sniffing.
A typical FTTB ISP providing services to people living in apartment buildings isn't doing a "carrier ethernet", I guess it's more like a datacenter. It just has a media converter connected to a switch or these days a switch with an SFP port per building and from there it's all regular copper and RJ45. In the end of 90s and early 2000s ISPs haven't yet figured out how to do it properly, how to isolate and authenticate customers, at the time there were plenty of hubs and dumb switches with plenty of sniffing going on, plenty of vulnerabilities in switches, a single customer was able to cause problems for the whole network.
The author writes on one reason why he released this open-access textbook in 2015:
> Perhaps the last straw, for me, was patent 8195571 for a roundabout method to force students to purchase textbooks. (A simpler strategy might be to include the price of the book in the course.) At some point, faculty have to be advocates for their students rather than, well, Hirudinea. [leeches]
BTW, the website for all versions (including PDF) of this text (currently version 2.0.2, September, 2020) is:
It's a shame that Sphinx, the documentation package that built the book, still doesn't support parts. Longer texts could benefit from parts, but the current developers are still not interested enough to allow it.
9 comments
[ 3.9 ms ] story [ 40.6 ms ] threadI don't think those were the reasons. The main reason I remember was cost. Shared media pretty much meant coax, and twisted pair is much cheaper. When switches became cheap enough and capable enough, the savings of running TP everywhere became the dominant factor. This also eliminated long cable runs and the frequent disruption of losing an entire segment at once, so reliability was also a factor, but IIRC those improvements were secondary in most purchasers' minds.
I guess cost reduction is a form of efficiency, but I certainly never heard privacy as a reason before.
> Perhaps the last straw, for me, was patent 8195571 for a roundabout method to force students to purchase textbooks. (A simpler strategy might be to include the price of the book in the course.) At some point, faculty have to be advocates for their students rather than, well, Hirudinea. [leeches]
BTW, the website for all versions (including PDF) of this text (currently version 2.0.2, September, 2020) is:
http://intronetworks.cs.luc.edu