Other than EVDO ("3G"), my only other connection options at my home are satellite (which sucks) and dialup. Ironically, I live on the hill above the most connec

Other than EVDO ("3G"), my only other connection options at my home are satellite (which sucks) and dialup. Ironically, I live on the hill above the most connected place on earth, the silicon valley. On the bright side, it's very beautiful up here. When I was healthy enough to tinker and test on my OS, the vast majority of my network transfers were due to downloading OpenBSD snapshots to test, binary packages, and updating CVS. Each snapshot is about ~280MB and depending on the arch/platform, they get rolled a couple times a week. I have a lot of different archs and I (used to) test on as many as time/health/bandwidth allowed. Unlike other open source projects, testing snapshots in OpenBSD is an important and continual process. Even when you're testing patches (i.e. compiling), the expectation is everything else should be at the current snapshot. The snapshots are essentially the baseline maintained so the patches/development are your only delta. BTW, the open source stuff is mostly FTP/HTTP transfers. Though I do have a bittorrent client installed, it maybe gets used two or three times a year, only for small files, and without upload reciprocity due to the caps. There's also very simple web browsing (without scripting by default). These days, most all of my time is spent reading, and a lot of the reading I do is keeping up with tech developments on the web. Once or twice a week, I might indulge in a conference/talk video, but if the file is too large, I won't bother. Also, about once a week I'll pull a (legal) anime video that's about 250MB. My total network transfer ranged between 5 and 8 GBytes a month when I was doing active testing, so by comparison to other home users, it's not that much. I seriously doubt I need to remind you but for the non-network folks, file sizes do not accurately reflect total network transfer sizes due to protocol overhead. The Verizon bandwidth caps are based on total network transfer. The most interesting, and sometimes frustrating, thing about the Verizon cellular network is the filtering. They do some fairly brutal things to reduce bandwidth usage and prevent "abuse" as per their unstated definition. For example, if too much ICMP traffic originates from your system to a particular host, it is silently blocked by their filters. Another example would be their handling of POP3/SMTP protocols. In minimal usage (i.e. checking mail six to twelve times a day), the Verizon filters insert delays. If mail is checked "too often" (i.e. idiots who configure their phones to check mail every 10 seconds), the destination address is blocked completely. Some of the things Verizon does in PPP are also wonky, but more or less it works well enough. The "excessive use" throttle (filter) on old "unlimited" accounts like mine only kicks in when a rolling average (three months I think) of usage continually exceeds the 5GByte cap. I haven't seen it happen in months but if memory serves, it gets limited to 128Kbit/s and only lasts until the end of the monthly billing cycle. Needless to say, Verizon refuses to divulge any information about their filtering rules/triggers but even a novice at network monitoring can see the results in action. The latency can be annoying on EVDO ("3G"). Part of it is caused by the nature of wireless/cellular communications crossing the distance, but the other part is caused by spectrum resource limits of the towers and the corresponding on/off dance the radio in the modem does. A good way to grasp the problem is, you use a lot more spectrum when you're talking on your cell phone than when it is sitting idle in standby mode. It's an understandable limit of the engineering/design; Cellular data networks are simply not designed for "always on" operation.