Statistics

I have a few side projects related to the BBS, such as testing call and download quality from the BBS, along with UUCP mail/news transfers, all of it over VoIP.

People say modems over VoIP don't work, I'm here to quantify if that's true or not

Call completion stats, ongoing

I grabbed a bunch of modems and made thousands of phone calls over VoIP to the BBS from a 486 test rig, using a few different analog telephone adapters. I verified the RTP/voice data was going out over the Internet to my ITSP, it wasn't just going back and forth on my LAN. I tested not only connecting to the BBS, but actual Ymodem downloads of a test file, usually 64 KB in size. Success is defined as a download successfully finishing.

There are several questions I'm trying to answer here:

  • Are there any brand/models of ATA and voice gateways, plus configs, that I can recommend to people or steer away from?
  • Why do Zmodem downloads completely fall apart over VoIP, whereas Ymodem downloads fail (my hypothesis is audio round trip latency)
  • Are there any brand/models of modems that work better/worse over VoIP?

    The data: ccr.txt

    UUCP / UUCICO stats

    In a similar vein, the BBS uses UUCP to transfer Internet email and newsgroups in and out over modem over VoIP calls. The BBS dials out to a Raspberry Pi running Taylor UUCP, using the UUCP "g" protocol which is different than X/Y/Zmodem. There's some MS-DOS based tools that feed in simulated health check emails and newsgroup posts so there's at least some data to process on each call. Here I'm mainly keeping historical data from the Linux side, with some quick and dirty ChatGPT-driven development scripts to pull data from mgetty and Taylor logs and shove them into sqlite.

    UUCICO data here


  • © Wann Networks 2024