The Serial Port is a virtual museum that highlights technology and people we find interesting from years gone by. We strive to preserve history of all types, both physical and digital as well as through the innumerable stories of the people that lived it.
Got some retro gear to donate? Get in touch with us over email (see below "For business inquiries") or www.serialport.org
The first ISP I used had a FreeBSD i386 server you’d dial into. Sign in with your username to get a shell, prefix your username with P to start a PPP session. It was a small local ISP and I guess they serviced all their users from that single server? Crazy.
A-law nets (eg T1) were US only everything had to fit around mu-law PCM (voice) here in Australia NZ and Europe. Always thought it odd the US did 1.544 instead of 2.0mb TDM (T vs E) links (Mu law slots were larger iirc)
Funnier fact on the "Pentium Bug" a year later after it became widely know it actually got referenced (intel got lampooned even) in pop culture in the Freakazoid cartoon as the entire origin of the main character, and the main villain. A strange bug in their "pentacle" chip that took a convoluted thing to actually activate the bug.
Even though NAT gets a bad rap, I do think it was (and to a certain extent, still is) necessary -- although the best case scenario would have been to move to IPv6 in the 90s, as an IT company or service provider, you don't really have much control over what your customers are running (or the capabilities of their software), you just need to make things work without it being extremely burdensome. As an example, Windows didn't even include native IPv6 support until Windows XP (although it was later added as an optional update to Win2k SP4). And to a certain extent, NAT has allowed people and device manufacturers to be somewhat lazy as far as private network security goes -- which I think is part of the main reason for the very slow IPv6 transition (along with the fact that many of the familiar IPv4-related protocols work very differently with IPv6, such ARP vs ND, DHCP vs DHCPv6/PD/SLAAC, etc). Complicated even more by the fact that many routers currently in use by ISPs and service providers can barely even hold a full IPv4 routing table, let alone IPv6 in addition to that. It has only been in the last few years that most major US ISPs have supported IPv6 provisioning for customers (although many consumer routers and gateways have IPv6 explicitly disabled by default).
I work for a major telco as an OSP engineer, and formerly a CATV engineer and manager for a mom & pop company. Practically all local telco's will still provide T1's regardless of your experience with that call. Now, instead of the old T-carrier frequency, many have switched to HDSL (synchronous DSL) to provide the circuit to a smartjack, which then converts to T carrier and then again to data/dsu/etc. I have a feeling that your local telco's rep might not have known her stuff. Where T-carrier circuits really suck is the loop charges. The further you are from the central office, and then again the central office from the PoP - you pay per mile. When my uncle started internet on his cable tv system fiber wasn't available, so I had to order T1's. We eventual got up to 4 T1's, which we had bonded to act like a single 6M circuit. It was 20 years ago, customer provided speeds were much lower than today. I think we offered 512Kb/128Kb at the time on those 4xT1 across 300-500 customers - which worked because not everyone was online at the same time, nor downloading at the same time.
I remember first hearing about this "wireless LAN" technology in the early 2000s. I initially scoffed at it and thought about it similar to how I think of IR-based communication that some laptops and portable devices had at the time. "There's no way it could be fast and reliable enough to actually be useful". We used to run cat5e everywhere to do local LAN parties when in high school, and although 802.11a & b PCI cards existed then, they kinda sucked, and you had to have a free PCI slot (almost everyone used desktops and carried them around). It wasn't until I got my first laptop in 2005 which had a builtin Atheros 802.11b/g chipset (and a desktop Pentium 4... man that thing was like a space heater.. lol) and then started using the beloved Linksys WRT54G(L) wireless routers that I started to change my mind.
We use console servers (like the Tripp-Lite B097-048) all the time for out-of-band connectivity to switches, PDUs, etc. But I suppose it really just is a full-blown Linux box with 48 serial ports.... never really thought about it being similar to a terminal server. Also, this series is great. I had always wondered how the early dialup and T1 infrastructure worked. As a sidenote: even 7 or 8 years ago, the company I work for was still using actual T1 lines from Cox Business for voice services (connected up to an Asterisk/FreePBX box via a PRI card iirc). Got to glimpse inside of the outside telco closet a couple of times to see the demarcation point and all of the line cards. Now all voice is done via VoIP, which is also probably a lot cheaper.
I worked at digitalNATION. We hosted dedicated servers and had an OC3 connection. We had tons of these. I was a web developer, but I remember we had so many servers, we had expanded into two other buildings nearby. For some odd reason, everyone had a NeXT desktop next to their regular computer. I only used mine for email.
My god you do such a good job at explaining the technical details of all this old infrastructure. Everything I’ve always wondered about these old systems is answered on this channel. Amazing work!
Greatly enjoyed this video, both for its very clear explanations and the amazing ingenuity on display at DEC - this is what you get when you have a company run by talented engineers, and sadly not so talented on the sales & marketing side.
Oh I remember retiring these when I was in High School/First year of college. I got a job for two summers in Chicago and the company had a mostly thicknet and thinnet based network powered by mainly by this series of cabletron repeaters. I basically spent my first summer working on the older stuff, and then slowly helping convert people over to the newer 10Base-T 3Com equipment that existed on 2 of our 4 floors using old Cat3. All 200+ staff sharing a blazing fast 1.44Mbps off a T1 (for my High School self that was an insane amount of speed). The following summer these were finally gone, we were converting from Novell IPX to TCP/IP based, and they had structural cabling all using the newest Cat5 cable and central patch panels in one computer room (amazingly new technology at the time). And at the core of it all I got to help setup the first Ethernet switch I ever saw. It was an 8-port 100Mbps switch. Each floors connections went to a separate 3com hub, then each uplinked to the switch, and then 2 x 100 Mbps FDDI ports formed a full-duplex fiber ring for all the Novell Netware servers. I think we had close to 16 GB of storage across all the servers combined....just massive amounts of space! I believe I was told the 8 port switch ran about $16-17k (for context I was paid $7/hr, minimum wage was $5.15/hr, my total college tuition per year was about $3.5k, and I was working to save up $750 to buy probably a 10 year old car....that switch would have paid for a car and 4 years of school & board with some to spare). By the time I left college you could get Cisco 3500 XL switch’s for about $3k. That’s the difference 4 years made in networking.
Oh wow that's a blast from the past! When I was in High School my home PC was a Ambra 486DX4-100 D4100 Blue Lightning w/540MB hard drive and 16MB Ram. It was quite an upgrade from our previous Myoda 386SX-25. The Pentium P60 had just become the big new thing pushing down 486 prices and with the decline of Ambra it was a good opportunity to pick up a new but now heavily discounted flagship Ambra box. The 386 got a MB/CPU swap to a P60 later and went off to college with my older brother, but I spent the next 4 years surfing the BBS on this Ambra 486DX4 on my 33.6 baud modem and playing Space Quest, Kings Quest, Need 4 Speed, etc games with its amazing VESA graphics speed. I think eventually I tossed in an Overdrive and extra ram I snagged from somewhere, but the custom riser made it pretty much impossible to reuse the chassis with any future board. I think around 1997 it was finally retired and replaced with a generic K6-133 for my parents to use, and I went off to college with a P5-166 I had saved to buy from Gateway 2000. To be honest it was still probably in my parents basement till about 10 years ago sitting in boxes of old NES, SNES, Sega Genesis, etc stuff.
2:58 If anything, seems more like the Cobalt Qube was a nod to the NeXTcube - which, as it happens, was the system Tim Berners-Lee used to create the World Wide Web to begin with. So, it would have been a bit odd for Cobalt to try and sue Apple when the Qube seems to have taken inspiration from the computer which the web was invented on.
i think it was just a slight print erroe on the board i think they was just trying to mark the non wide white side as the positive but msitakes happenw ith older stuff
ha ha ha yea right, apple invented wifi. ha ha ha ha ha ha ha they changed nothing. They just copy off everybody else, just about 8 years after everybody else already has the tech.
I hope you get that PAS16+ fixed, I bet it got killed the same way mine did, by connecting it directly to non-amplified speakers. Tried to have mine fixed without any luck but that attempt was not with an electronics repair expert in any case so there is hope!
I had the pleasure of Owning a Media Vision Pro Audio Spectrum 16+, that had the best high freq. response of any card I had before or up to the one included on the ASUS NForce Deluxe Gold motherboard (best onboard audio ever with 3DSPs) the bass needed a bit of pull up but was good regardless. I also had the pain and sadness of killing the card as I was connecting it directly to 2x15W 4ohm speakers which ended up killing its pre-amp. I regret that loss to this day. Still, my favorite card ended up being the AWE32 because of the onboard RAM and being able to have it play modules while I was programming in turbo pascal in the background barely using any CPU or RAM from the main system. And good audio quality too so yeah, best toy ever audio wise at the time :P
You think power supplies are bad? You just wait until you get a discharge from the main capacitors powering a CRT Electron cannon then you come back to me...
Incredible, that motherboard has a decent VGA chipset on itself, I hope they used the Tseng Labs for 2D VGA and on the VLB card they only used the 3D part of it, this is the best performing configuration assuming we can use it that way.
That VLB card is an awkward poorly designed one for one single reason, the base OAK/OTI VGA chipset is a slow snail, the 3D accelerator chip aint too bad though assuming you actually manage to get support for it that is...
I loved these servers. I deployed several e-commerce sites on them in 1998 and 1999 and they were about as simple to setup, administer and develop on as you could possibly want. I used several different front ends for them that I can’t recall the name of now, but basically just like CPanel and others are today. Amazing how in some ways things really haven’t changed all that much.
I ran my own ISP from 2013-2018. I bought space in a datacenter that had direct links to larger ISPs. I was able to buy several high bandwidth uplinks to several of these ISPs, let’s say the uplinks were aggregated via a load balancer and the overall bandwidth was 100Gb/s and it cost me $100k/mo, if I had 100x1Gb links it would cost way more than that. When a customer signed up I would lease them a router capable of accepting a tunnel and talking BGP/MPLs, and I would broker a deal between the customer and local ISP to use their infrastructure to connect to my datacenter which I’d get a kickback for and/or a deeply discounted rate. The purpose of this ISP was to provide an IP telephony service where their signal processing, and accounting was centralized as a value add, and could be encrypted end to end at a higher cost. I would also lease them phones and PoE switches. If the customer had multiple locations I would use my datacenter as a hub and keep “local” calls off the public internet. I would keep each tenants traffic separate by accepting their connections into a firewall which was the tunnel initiator, and then convert their traffic on the other end to a private VRF which kept traffic inside the DMZ logically separate. Ignoring the fees I’d charge to lease the equipment, with that 100Gb/s connection I could support 1,000 customers at 100Mb/s for $1,200/mo which would generate $1,2000,000 a month hypothetically on top of equipment leasing fees, which more than doubled revenue. Those 1,000 customers could easily have 30,000 handsets which only cost me $150 a piece and each handset came with a $35 fee per month with a 12-36 month contract which adds up to $1,050,000/month which would net me $675,000/mo. Oh and if you want me to send someone out to set up and register the phones it’s and hourly rate, or I could use free-setup as a loss leader. It was like a cheat code for printing money. The only thing is the backend setup is a pain, and administration sucked (eg setting up new extensions, users, training, directories, porting numbers, toll free, e911). The total revenue for my example would be over $2,000,000/mo, the net would be something like 12% after datacenter fees, the fees paid back to the local ISPs, payments on capital, administrative labor, installation labor (we did not provide cabling services, but that could’ve been a big opportunity). We also had to lease our own vehicles because we were at customer sites all the time and reimbursing mileage is a pain.
