Jump to content
IGNORED

How did dial-up internet work?


Recommended Posts

2 hours ago, bluejay said:

If I understood correctly, the internet service provider would encode a website using a file transfer protocol into data packets that a computer can receive through a modem, and some software on the computer would decode the data packets sent by the ISP and display them on the web browser?

 

I did some googling about all these networking protocols and was bombarded with a 7 layer abomination of what seemed like a million different protocols that could be used for all sorts of... anything. I think it might be a while until I understand truly how these work.

 

Anyhoo, many thanks to everyone for your responses.

You've got Internet IP packets. Those are the "Internet stuff."

 

To get this Internet stuff moving between computers, you must carry them in something. It could be your Ethernet, modem, cable box, whatever. Actually nowadays you can even transport them via Bluetooth. Note that all these transport methods can also carry stuff that has nothing to do with the Internet, like the native files and data of the respective computer systems.

 

The Internet apps and servers (software servers) don't care how they access the IP packets as long as they eventually get the data and interact with it.

  • Like 1
Link to comment
Share on other sites

20 hours ago, Krebizfan said:

Consumer use for Compuserve started in 1979. Off-peak hours were cheaper. I forget exactly when the free starter kits emerged in an attempt to attract the more casual user as opposed the big clients that used the service during standard business hours. 

 

A classmate of mine received such a kit with his new 300 baud(!) modem in about 1987. If memory serves, the offer had already expired by the time he got it, so presumably they were first being distributed a few years prior.  

Link to comment
Share on other sites

On 5/9/2021 at 8:33 AM, bluejay said:

I've thought this through, and it's occurred to me that for a website to be able to be directly accessed via phone, the computer would somehow have to know the phone number of "www.google.com" and have the modem dial it. This doesn't seem to make a whole lot of sense. How did dial-up internet actually work?

Simplest explanation..    You would dial up your internet provider (ISP),  that was often some local company with local phone number (you paid for long distance calls back then). or something like AOL

 

Then you and your ISP would use a protocol known is PPP or SLIP to communicate.    This basically encapsulated the internet protocol aka TCP/IP..   you got an IP address and to all internet-aware applications, you were connected   These applications don't care about how you got connected... could be ethernet, phone line, satellite or even sending packets via pigeon,  as long as you have an IP address and a way to route packets through a gateway, they could do the rest.

 

Finding google was just doing a DNS lookup for www.google.com and optaining an IP for google, same as it is now.    The dial-up modem just made it slower.

 

EDIT:   I will add that it was kind of mind-blowing when I first got a PPP connection, and got it working (it wasn't self-explanatory in 94, a lot of the terminology was new to me)  Before that direct internet access and IP networking was something only large organizations could afford.   Now most of us have IP networks in our homes!   Then another mind-blowing moment was when someone told me I could download Netscape Navigator (early web browser) completely for free and run it with my PPP connection.  Suddenly I was able to browse the graphical web from home!

Edited by zzip
  • Like 3
Link to comment
Share on other sites

Internet access works the same no matter what.  On a dial-up modem, you use an app that dials the ISP's number and logs in with your username & password through regular phone lines.

 

With broadband your DSL or cable modem does all of that automatcially over seperate signals carried over phone or cable TV lines, and far faster as well. Same with smartphones that use 4G/5G signals...

 

 

  • Like 1
Link to comment
Share on other sites

well.. yes and no.

 

dialup modems just create the physical layer. Like all network transports, it is not really accurate to call it the physical layer though.

 

The modem pair create a sychronous communication channel over the telephone lines, with added error correction and recovery, at a negotiated line speed. This is a 1:1 pairing with a remote modem.  They do this by modulating a carrier frequency using a known shared modulation schema. (this is where different modem flavors play in.)

 

With a BBS, this is just fine:  It works just like a serial connection after being established, with the modems doing the translation to and from the modulation schema transparently to the user and any connected applications.

 

For dialup internet however, actual IP protocol is routed on top of another protocol, the Point to Point Protocol, or PPP.  This is because IP protocol is not really intended for 1:1 connections.  The PPP transport functions kinda-sorta like a network switch, and restricts traffic not intended for that channel from traversing that channel, while also assuring that actual IP style network clients upstream can get all the data transmitted from the PPP connection, like IP networks expect, and also serving some 'proxy server' like functions.

 

PPP can/could be used in this way with any 1:1 paired transport, not just dialup modems. That included direct serial, microwave dish uplinks, et al.

 

 

It is important to point out that very early in the history of the consumer internet, PPP and IP networking stacks were **NOT** part of the operating system. Microsoft's OSes did not have native functions to handle modems or internet connections built in until windows 95, and then they were still very immature.

 

In the win3x days, you had to use a 3rd party dialer app, and have some other goodies loaded and set up, to use a dialup modem with your free copy of netscape. :)

 

A very commonly used one was WinSock PPP.

 

https://org.coloradomesa.edu/~jerry/systeminfo/ppp.win31.html

http://www.steptail.com/guides:connecting_windows_3.1_to_the_internet

 

You started winsock PPP first, and it handled your modem, dialed your ISP, established the PPP connection, got you an IP address, and then "you were on the internet", but you still had to open your browser after the fact.

 

Later versions of Netscape integrated the functions of Winsock PPP and pals, so they could dial all by themselves- but that quickly become obsoleted and culled, once microsoft started integrating dialup networking components into the OS-- the browser could simply tell the OS to "dial this connection for me", and the OS would handle it. (sometimes.)

 

By the end of the 90s, and early 2000s, ADSL and cable modems were starting to become a thing, and always-on internet was becoming a reality.  Such functions as dialup networking are now legacy footnotes in the annals of internet history.

 

 

On the mac side of things-- you had to use applications similar to winsock ppp with OS system versions 6.x, but had a PPP os extension for system 7.

 

https://www.radford.edu/~helpdesk/ppp/Macintosh/index-7.5.html

https://kb.iu.edu/d/adaq

 

  Netscape navigator and communicator 3 could latch onto that PPP extension, and dial for you.  The PPP extension wanted to know what mac serial port you were using, or-- if you were so fortunate-- what modem addin card you had-- what your ISP's username and password were (needed to negotiate PPP), what the number to dial was, what line speed to use, etc.

 

Mac serial ports of the time were proprietary things, so you had to have either special adapters, or special mac-oriented modems.

mac-serial-ports.jpg?w=224&ssl=1

 

 

 

 

 

 

Edited by wierd_w
  • Like 5
Link to comment
Share on other sites

13 hours ago, wierd_w said:

It is important to point out that very early in the history of the consumer internet, PPP and IP networking stacks were **NOT** part of the operating system. Microsoft's OSes did not have native functions to handle modems or internet connections built in until windows 95, and then they were still very immature.

Even Windows 95 didn't install it by default.  (at least not vanilla, pre-Service Pack Win95).    It was on the CD, but you had to know which components to add to make it work.

 

Linux distributions even back then came with all the tools to initiate PPP/Slip connections, but it was far from self-exploratory how to run it.

 

When I got my first Internet dial-up,  I was still on Atari ST, but was in the process of looking into getting a PC.

I used KA9Q on the Atari ST-   this was pretty limited in what it could do, but I think I mostly used it for FTP.   When I got my PC built, I was using Trumpet Winsock on Windows 3.1,  and eventually got pppd working correctly on (Slackware) Linux.

  • Like 2
Link to comment
Share on other sites

  • 2 months later...

this brings back really bad memories.

 

Basically, your computer would connect to the internet, you'd get a busy signal and SLOWLY connect to your desired website. Sometimes it wouldn't even connect. It did the same tone every time you needed to connect.

 

We didn't even have a 56k modem. It was just that cursed dial up for months until the 2000's. I HATE dial up

  • Like 2
Link to comment
Share on other sites

This internet thing seemed fundamentally different to me than direct dial to a BBS. Complex. Mysterious.

 

By the late 80's the "traditional" old-hat stuff I was doing on the Apple II like WarEz were dying. AE Lines long gone. GBBS, TeleCat II, Networks BBS.. All faded glory. And all of it built around 300/1200/2400 baud speeds via direct computer-to-computer, 1-1, modem-to-modem connection. Web pages were still years in the future.

 

Early dial-up internet with America Online wasn't too bad. But I had a 14.4K modem to start with, and 56K soon thereafter.

 

At first I sorely needed AOL's training wheels. By 1995-1997 or thereabouts I got a real browser like Netscape Navigator, learned the significance of Mosaic. Learned SIGs and Newsgroups and IRC. Then fell into the grind like everyone else.

 

I stopped paying attention to speeds when broadband came out. It was lightyears faster than any dial-up imaginable.

Edited by Keatah
Punctuation perfection.
  • Like 2
Link to comment
Share on other sites

Totally. All those needless APIs, libraries, Java, the need to animate crap. And all the middlemen and advertising and two-way communication. Excessive graphical images. Cookies. Security. All that crap. Totally unnecessary. Glad I'm not part of that rat race. Fuhk'emall!

 

A simple page with a banner and buttons to indicate what you're looking for is more than sufficient. And text will do. No animation needed.

  • Like 1
Link to comment
Share on other sites

21 hours ago, OLD CS1 said:

I hate ComCast more. If I ever move to an area where my only choice for broadband is ComCast, I will happily go back to dial-up.

Cox: the cable company whose name is an onomatopoeic representation of the quality of their service.

  • Haha 1
Link to comment
Share on other sites

There's been talk elsewhere in the thread about what happened when you dialled in to one of your ISP's numbers, but a few key parts haven't yet been mentioned.  So, to add that information in (and it has been a very long time since I've had to work with these technologies; additions or corrections appreciated):

 

The number that your modem would dial was known as a PoP, standing for Point Of Presence.  This is not to be confused with the POP3 email protocol, which has no bearing at all on the subject at hand :grin:

 

PoPs could be hosted directly with your ISP (e.g., in a modem bank) or by a third party.  Third parties were typically carrier-grade providers such as Level 3 or MCI; there were others, but their names escape me right now and most of them were absorbed into other companies over the years.

 

In general, it was preferable to be able to dial in to a modem located at your ISP since it was the most direct connection to the Internet you could possibly establish.  However, as others have mentioned, calls weren't flat-rated at the time as they effectively are now, and dialling long-distance to access a PoP got expensive real fast.  This is why third parties provided PoPs: they could place them where the ISP couldn't, thus providing service at local call rates at great distance from the ISP's location.  It also brought about the creation of ISPs that were able to operate on a national level.

 

Once you were connected to the POP, a PPP (Point-to-Point Protocol) session would be established between your PC and the POP.  At this point, authentication would take place.  This was typically handled via RADIUS (Remote Authentication Dial In User Service), which would verify the credentials being presented and either complete establishing the connection to the ISP or reject the connection attempt and hang up.

 

A note about PoPs operated by third parties: they were likely to be shared by several ISPs.  Because of this, there needed to be a way to let RADIUS know which ISP your account was with.  The usual way of handling this was by adding a prefix to the username that RADIUS could use to identify which ISP it needed to query against - so if your regular username was 'jrandomcustomer' and you had an account with EarthLink, you'd use 'ELN/jrandomcustomer' as your username when logging in to the PoP.

 

And, yep, "logging into the PoP," is a correct statement.  Your credentials were provided to the RADIUS instance running at the carrier providing the PoP, who then verified them on the backend against the RADIUS instance at your ISP.  In a sense, the carrier was acting as a broker between you and the ISP and determining if you were permitted to be online or not - after all, they owned the lines providing the connectivity from the PoP back to the ISP, so had a vested interest in only letting authorised users on.

 

Finally, once authentication has been successful, the carrier routes your traffic to your ISP, who then put it out onto the Internet for you.  Won't get into the 7-layer burrito that is the OSI model, but this is the point where the actual browsing, emailing, FTPing, and Usenetting of the day took place.

 

A couple of notes:

 

Because PoPs were shared amongst different ISPs, busy signals weren't uncommon when dialling into them.  Nights were the worst: everyone would come home from work and immediately go looking for porn online.  Some ISPs and/or carriers combatted this by dropping sessions after a couple of hours, regardless of whether or not they were active.

 

One additional reason for not wanting to use a long-distance PoP: the call quality of analogue (aka POTS) phone lines typically degraded based on distance.  Dialling into a PoP on the other side of town was generally a better bet than trying to reach your ISP's own in-house PoP 1500 miles away as a result.

 

It was very much possible for RADIUS instances on individual PoPs to fail.  Sometimes it could be just one ISP that was affected, sometimes the entire PoP.  This naturally led to congestion on other PoPs in the area as people would start dialling different numbers than they usually would to get online.  Although a PoP could normally accommodate more than one call at a time to it, there were limits.

 

There were two instances where I got to witness a RADIUS failure that left RADIUS wide open to anyone who dialled up to the PoP and had a terminal window open.  It was fun for a bit, but I eventually called the ISP in question and let them know that they needed to have a word with the carrier.

 

 

  • Like 1
Link to comment
Share on other sites

On 7/17/2021 at 8:20 PM, Gamemoose said:

What was nice with the dial up era Web was that many pages were designed to be speedy (less flashy stuff-got right to the content). With broadband came eventual bloat that makes some sites slow if your speed isn’t where it should be that day.

True and browsers were also designed to load the text first, then the images.    The good browsers checked the size of the images first and reserved the correct amount of space in the page for them, they would also load in a low res version of the image first and replace it with a higher-res when everything else was done loading.

 

The bad browsers didn't know the size of the image, and as a result would shift the text around as the images loaded...  that was highly distracting when you were trying to read something!

  • Like 2
Link to comment
Share on other sites

On 7/18/2021 at 3:02 AM, wierd_w said:

It's still in most linux distros.  There is also a windows port, and there is a fork for dos.

Indeed, and I also have it on my Solaris machines.  But there are not a lot of sites which work very well with it, which, I guess I should have said more specifically, is more my lament.  Thus, not so much that I miss its presence but rather that I miss its full usefulness.

  • Like 1
Link to comment
Share on other sites

On 7/18/2021 at 11:32 AM, x=usr(1536) said:

There's been talk elsewhere in the thread about what happened when you dialled in to one of your ISP's numbers, but a few key parts haven't yet been mentioned.  So, to add that information in (and it has been a very long time since I've had to work with these technologies; additions or corrections appreciated):

 

The number that your modem would dial was known as a PoP, standing for Point Of Presence.  This is not to be confused with the POP3 email protocol, which has no bearing at all on the subject at hand :grin:

 

PoPs could be hosted directly with your ISP (e.g., in a modem bank) or by a third party.  Third parties were typically carrier-grade providers such as Level 3 or MCI; there were others, but their names escape me right now and most of them were absorbed into other companies over the years.

 

In general, it was preferable to be able to dial in to a modem located at your ISP since it was the most direct connection to the Internet you could possibly establish.  However, as others have mentioned, calls weren't flat-rated at the time as they effectively are now, and dialling long-distance to access a PoP got expensive real fast.  This is why third parties provided PoPs: they could place them where the ISP couldn't, thus providing service at local call rates at great distance from the ISP's location.  It also brought about the creation of ISPs that were able to operate on a national level.

 

Once you were connected to the POP, a PPP (Point-to-Point Protocol) session would be established between your PC and the POP.  At this point, authentication would take place.  This was typically handled via RADIUS (Remote Authentication Dial In User Service), which would verify the credentials being presented and either complete establishing the connection to the ISP or reject the connection attempt and hang up.

 

A note about PoPs operated by third parties: they were likely to be shared by several ISPs.  Because of this, there needed to be a way to let RADIUS know which ISP your account was with.  The usual way of handling this was by adding a prefix to the username that RADIUS could use to identify which ISP it needed to query against - so if your regular username was 'jrandomcustomer' and you had an account with EarthLink, you'd use 'ELN/jrandomcustomer' as your username when logging in to the PoP.

 

And, yep, "logging into the PoP," is a correct statement.  Your credentials were provided to the RADIUS instance running at the carrier providing the PoP, who then verified them on the backend against the RADIUS instance at your ISP.  In a sense, the carrier was acting as a broker between you and the ISP and determining if you were permitted to be online or not - after all, they owned the lines providing the connectivity from the PoP back to the ISP, so had a vested interest in only letting authorised users on.

 

Finally, once authentication has been successful, the carrier routes your traffic to your ISP, who then put it out onto the Internet for you.  Won't get into the 7-layer burrito that is the OSI model, but this is the point where the actual browsing, emailing, FTPing, and Usenetting of the day took place.

 

A couple of notes:

 

Because PoPs were shared amongst different ISPs, busy signals weren't uncommon when dialling into them.  Nights were the worst: everyone would come home from work and immediately go looking for porn online.  Some ISPs and/or carriers combatted this by dropping sessions after a couple of hours, regardless of whether or not they were active.

 

One additional reason for not wanting to use a long-distance PoP: the call quality of analogue (aka POTS) phone lines typically degraded based on distance.  Dialling into a PoP on the other side of town was generally a better bet than trying to reach your ISP's own in-house PoP 1500 miles away as a result.

 

It was very much possible for RADIUS instances on individual PoPs to fail.  Sometimes it could be just one ISP that was affected, sometimes the entire PoP.  This naturally led to congestion on other PoPs in the area as people would start dialling different numbers than they usually would to get online.  Although a PoP could normally accommodate more than one call at a time to it, there were limits.

 

There were two instances where I got to witness a RADIUS failure that left RADIUS wide open to anyone who dialled up to the PoP and had a terminal window open.  It was fun for a bit, but I eventually called the ISP in question and let them know that they needed to have a word with the carrier.

 

 

Oh man…I remember the modem bank room at one of my old employers. It was a couple shelves with US Robotics modems (don’t recall the model). You saw all these red lights glaring at you in the dark when you opened the door.

 

It took a bit for me to find a working local number to call for general Web access after I wanted to jump off AOL. Connection speed varied but got better when I worked for the aforementioned employer and got service for free :) .

  • Like 1
Link to comment
Share on other sites

43 minutes ago, Gamemoose said:

Oh man…I remember the modem bank room at one of my old employers. It was a couple shelves with US Robotics modems (don’t recall the model). You saw all these red lights glaring at you in the dark when you opened the door.

My company used the Ascend Max, like this one.

 

https://www.recycledgoods.com/ascend-communications-inc-mxhp-4t1-max-4000-router-0700-0095-001/

 

We had PRI (digital voice over copper T1, 23 64k ISDN B-channels and one D-channel used for signalling,) cards in ours, but ISTR we had the option for RS-232 cards which could support a certain number of analogue modems in each.  Later the Ascend TNT came into play with a T3 instead of the old copper PRIs (IIRC, it was actually many PRIs over the T3 fiber.)  I started with 56k at home and moved to 128k ISDN (two bonded 64k ISDN channels) over an Ascend Pipeline 75.

 

At one point the company took over a local ISP which used a Livingston Portmaster 2e, so we also inherited a bunch of 56k modems (of which I still have about a dozen, but the PM is long dead and gone.)

  • Like 1
Link to comment
Share on other sites

2 hours ago, Gamemoose said:

Oh man…I remember the modem bank room at one of my old employers. It was a couple shelves with US Robotics modems (don’t recall the model). You saw all these red lights glaring at you in the dark when you opened the door.

Likely it was this one or perhaps one of its immediate predecessors.

https://www.usr.com/products/56k-dialup-modem/usr3453c/

 

And while they don't make that exact model today, they do make a very similar one.

https://www.usr.com/products/56k-dialup-modem/usr5686g-pro/

  • Like 1
Link to comment
Share on other sites

57 minutes ago, OLD CS1 said:

Oh, man, I remember those.  Worked at a couple places in the late '90s / early 2000s that used them.  From what I recall, they were pretty reliable.

58 minutes ago, OLD CS1 said:

We had PRI (digital voice over copper T1, 23 64k ISDN B-channels and one D-channel used for signalling,) cards in ours, but ISTR we had the option for RS-232 cards which could support a certain number of analogue modems in each.  Later the Ascend TNT came into play with a T3 instead of the old copper PRIs (IIRC, it was actually many PRIs over the T3 fiber.)  I started with 56k at home and moved to 128k ISDN (two bonded 64k ISDN channels) over an Ascend Pipeline 75.

Somewhere around here I might still have a box of about 3 dozen CSU/DSU cards pulled from various Cisco devices.  I do not miss those days at all - partial PRI failures were the bane of my existence with a certain PBX I was once responsible for.

1 hour ago, OLD CS1 said:

At one point the company took over a local ISP which used a Livingston Portmaster 2e, so we also inherited a bunch of 56k modems (of which I still have about a dozen, but the PM is long dead and gone.)

Still have a few of these kicking around for those couple of times a year when I get an itch to do some analogue telephony over VoIP ?

 

IMG_0202.thumb.jpg.e8cf4b13d47d8bb12802800875262b7c.jpg

  • Like 2
Link to comment
Share on other sites

6 hours ago, Keatah said:

Likely it was this one or perhaps one of its immediate predecessors.

https://www.usr.com/products/56k-dialup-modem/usr3453c/

 

And while they don't make that exact model today, they do make a very similar one.

https://www.usr.com/products/56k-dialup-modem/usr5686g-pro/

The company we picked up used a shyte-tonne of Hayes and Supra 28.8 modems.  They were going to have to upgrade from analogue lines to PRIs to support 56k.  The cost to upgrade the analogue lines to 56k would have cost quite a lot: the only options for direct modem replacements to 56k on the dial-up side was X2 and then only with other X2 modems.  v.90 was just around the corner and the Ascend Max offered 56KFlex with software upgrade to v.90 (and eventually v.92, but I am pretty certain we never got that far.)

 

We had a remote PoP run on USR equipment (Total Command?,) but it was horrible.  Later we bought services from a company which provided PoPs all over the US (SuperPOPs, I think it was called.)  It used a customized version of the Windows dialup from the IE IAK and was pretty slick.

  • Like 2
Link to comment
Share on other sites

On 7/19/2021 at 11:56 PM, OLD CS1 said:

We had a remote PoP run on USR equipment (Total Command?,) but it was horrible.  Later we bought services from a company which provided PoPs all over the US (SuperPOPs, I think it was called.)  It used a customized version of the Windows dialup from the IE IAK and was pretty slick.

I was reminded of having been called out to service one of these types of PoPs about 10, maybe 12 years ago.  What I found was these companies would co-locate their dial-up equipment wherever they could.  Kind-of how cellular companies will lease property from an owner (my aunt hosted a BellSouth tower in the corner of her land,) these dial-up companies would rent a room (or corner of a room) from some business where fiber or ATM-over-copper connection was available.  The one I serviced was in the stock room of a liquor store.  The PoP consisted of an Eton UPS, a fiber DMARC , and fiber connection to what I think was a Cisco box, and could service 184 sessions (four dual PRI cards.)  That was it.  This one was a mini-rack mounted on a wall, but I have seen ones with a full rack and batteries and such.

 

I also saw one in the back of a craft store, equipment hanging from its wires and covered in dust.  But running.

  • Like 3
Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

Loading...
  • Recently Browsing   0 members

    • No registered users viewing this page.
×
×
  • Create New...