From my archives... this is what the command center of a mid-1990s dialup ISP looked like.
Later iterations would be more advanced, but this is how it started. A natural progression and leap of faith from the multi-line BBS it replaced.
From my archives... this is what the command center of a mid-1990s dialup ISP looked like.
Later iterations would be more advanced, but this is how it started. A natural progression and leap of faith from the multi-line BBS it replaced.
This is making me so nostalgic for the first days of our little ISP. Looks very different and yet the same. Even the fans blowing onto the shelves of modems to keep them cooler! We had put ours on metal rack shelving in the middle of the room so the air could circulate more freely.
Are the 3 chassis on the desk the terminal servers for the modems? I can't make out the logo or brand, and I no longer remember what we were using at this stage.
@autumn How oversubscribed was each phone line?
Busy signals were rare even at my small town, rural ISP and Iβve long wondered how many lines they had for their userbase.
@amd There were two kinds of oversubscription to watch out for:
1) The ratio between customers and number of phone lines. This is where you'd get busy signals if too many people tried to be online at once.
2) The ratio between the theoretical max bandwidth demand of all the modems, and the size of the upstream pipes to the Internet. Get this wrong, and everybody starts getting slower-than-modem-speed throughput.
I can't remember exact numbers for either of those, but it was something we monitored and watched closely. Lots and lots of graphs. Usually we could keep well ahead of both, but it took some careful anticipation, given how long it took for the phone company to get new circuits installed.
We used to do log analysis to track the daily "high-water mark" for each pool of modems, and plan based on its dynamic ratio to our user count, so at peak usage there should be just a few lines free.
We figured out early on that if people never got busies calling in, they'd hang up when not using it. If they got busy signals more than very rarely, they'd start camping on the lines to avoid it, causing a vicious circle, so this "overprovisioning" actually reduced usage.
While I'm reminiscing, we also - very rarely - ran a waitlist for new users when we had trouble getting lines installed, which unexpectedly had the opposite effect than we'd expected.
It was the "Studio 54" effect - if a long line of people are waiting to be let in, you assume it must be good! The local word of mouth became "LavaNet is so obsessed with service quality they put users on a waitlist rather than get busy signals", and we suddenly got a ton of new signups.
Beautiful photo, if anything because of the memories it recalls.
I worked in a room very much like that in 1998, before we got the Ascends working.
Good ole Livingston portmasters!!!
@autumn ooooooo we had those exact livingston portmaster 3e's! our setup was very, very similar at the isp i worked at when i was a kid.
gotta love the fan cooling off all those couriers π
@autumn
Oh how I wish I had pictures of the community dialup system I set up around that time! The power bar tandeming was outrageous. Shelves full of 56k modems, aggregated through a long forgotten device connected to a (gasp!) 56k telco data line.
The excitement when we eventually upgraded to a full T1 was unimaginable.