I worked I a computer room from 1989. I worked in the tape pool. We watched a screen, noted the tapes the system was requesting, located the tape, and mounted on a drive.
I think you might be surprised. I read recently that most of the banks still use software that was first developed in the 1960s. They have more recent systems as well, but they're all built on top of the older software, like layers. They can't get rid of the old software because that would mean shutting down all the systems for a certain period of time and that isn't an option since they have to be "on" all the time. So you've got software from the 60s, 70s, 80s, 90s, etc. still being used, as well as new systems. Apparently this is one reason why when you put money in your bank it can still sometimes take a few days to show up in your account.
I read somewhere that in the run up to the year 2000, the year of the anticipated millennium bug, the UK government had regular meetings with British banks to see how prepared they were. Turns out that as recently as 1999 banks were still calculating transactions in £/s/d and then converting to £/p !!! :O
After 9/11 the brokerage/bank I worked for began the process of converting to a virtual machine. They got rid of over 100,000 magnetic tapes. Once they copied all the tapes they needed they got rid of the tape pool operators. I stayed behind and worked for several more years.
@@ajs41 My uncle confirmed what you just said. He has been coding in assembly language for IBM mainframes since the 1970's and is still in demand for exactly the reasons stated- the modern abstraction layers and hardware still (at the central core) run software code from decades before. He said that trying to change over completely to new systems is like trying to change all four tires on a race car while driving over one hundred miles an hour! You have to completely stop to allow the change over to take place.
At 0:45 he points out the OCR characters at the bottom of the check. If the machine can't read them because they are smudged or the check is torn they are routed to another machine that can read them. I interviewed to repair that machine at my first job interview after the Navy. It was called the Lundy--Mae and it was made by Lundy Farringrton it had four hundred and forty circuit cards. Fortunately, I did not get the job because it only paid about $10,500 a year. This was in 1979 but even then I would have struggled in NYC on that income.
"Fortunately, I did not get the job" So why did you apply and waste the interviewer's time? Particularly if you might have got the job whilst not wanting it whereas another candidate was desperate for the job.
9:46 those auto-loading tape drives are always fun to watch. Too bad they cut away without showing the tape loading the nightly batch processing is why banks, at least in Denmark, don't process payments and cash withdrawals between 23.00 and 06.00 on weekdays, or from Friday 23.00 to Monday 06.00. So if you for instance use an ATM to pull out money over night, the bank won't know till the next morning, and that can easily be abused to overdraft your account by far more than the bank allows. For the first 10 years or so with debit cards the bank owned ATMs didn't work at all at night because of that, but since the mid-1990s or so the transactions are still stored even when the bank is closed, but it's not processed till the next morning
Fun fact: The same software and methods are still in use with some banks. At least the hardware changed. But the IT of most banks is an horrible mix of ancient systems.
Why wouldn't each local office instead: 1. manage local universal identification numbers (LUIDs). 2. print out and mail the bank statements themselves? 3. transmit the transactions and new customer CRUDs back to the head office purely for archival purposes?
firstly because the computers were super expensive and required a ton of space (and electricity), so it was unrealistic for every branch to have their own computer. The computer they show in this film would've cost a couple million £ back then and took up most of a floor in a building. Secondly your suggestion would mean that each branch would not be notified if a customer made transactions through a different branch. All the data is sent to the head office specifically so all transactions from all their branches are processed together so each branch can work from identical data. Due to the limitations of how the computers worked back then and the total lack of internet (or secure communication), this was the most efficient way of doing it. The computer was purely used for processing transactions, nothing else. The local branches didn't have any computers whatsoever and did everything on paper, so if you went into the bank in the morning and withdrew some money, and then came back later to ask about your current balance, they could only tell you the balance from that morning because your withdrawal wouldn't get processed till after the bank closed. Bank branches didn't get their own computer systems till the mid-1980s, and even then it only served to cache the data sent to and received from the head office, but transactions are to this day only processed in one central location. Banks still largely function according to these old principles
@@thesteelrodent1796 Awesome answer. Thank you. Now I'm wondering about a failsafe against fraud: what if, for example, five people withdrew $100 from an account that only had $300? They were at separate branches and made the withdrawals within hours of each other.
@@thesteelrodent1796When I worked at the bank, updated customer account records were received in the morning by microfilm to be read on the microfiche monitor. Also received were unreadable cheques and credits from the clearing house for manual branch input. We had a Burroughs TC500 with paper tape programming and modem. In early 80s we moved to Burroughs Modular Terminals with the luxury of a monitor and 5.25" floppy drives.
so.... the , cash transaction takes place, while that cheque waits at that table to get processed later, right? i'd like to withdraw a fantazillion quid pleace... and would you mind me, waiting right here, at that desk?... you don't happen to have a pen, you could lend me a minute there, while you get my cash?
right. You give the girl a cheque and she writes down that she received it. The information from the cheque is sent to the head office and the cheque itself is sent to the issuer's bank so they can process it as well, but because it's a cheque the amount on the cheque will be on hold until the issuing bank transfer the money to your bank. This is why cheques took several days to clear, and only when your bank has processed that they received the money will it be added to your balance. But they won't let you withdraw more than the current balance - or available amount if you have a credit line - because they know what your balance was the day before.
Good video, but quite a robotic description of how primitive data bases work. No real explanation of field names, and how it relates to the entire data set.
3:24 quite possibly the clearest spoken Scottish in modern history
"MUD'RR"
I worked I a computer room from 1989. I worked in the tape pool. We watched a screen, noted the tapes the system was requesting, located the tape, and mounted on a drive.
That's a job I wanted to do. However, by the 90s companies really didn't use reel tapes anymore. They used carts.
That squelchy synth sequence at the start is clearly an embryonic form of acid house.
Fascinating look into the past of high end data processing. I cant imagine modern banking systems are handled quite like this anymore. :)
I think you might be surprised. I read recently that most of the banks still use software that was first developed in the 1960s. They have more recent systems as well, but they're all built on top of the older software, like layers. They can't get rid of the old software because that would mean shutting down all the systems for a certain period of time and that isn't an option since they have to be "on" all the time. So you've got software from the 60s, 70s, 80s, 90s, etc. still being used, as well as new systems. Apparently this is one reason why when you put money in your bank it can still sometimes take a few days to show up in your account.
I read somewhere that in the run up to the year 2000, the year of the anticipated millennium bug, the UK government had regular meetings with British banks to see how prepared they were. Turns out that as recently as 1999 banks were still calculating transactions in £/s/d and then converting to £/p !!! :O
After 9/11 the brokerage/bank I worked for began the process of converting to a virtual machine. They got rid of over 100,000 magnetic tapes. Once they copied all the tapes they needed they got rid of the tape pool operators. I stayed behind and worked for several more years.
You mean, repeatably and reliably?
@@ajs41 My uncle confirmed what you just said. He has been coding in assembly language for IBM mainframes since the 1970's and is still in demand for exactly the reasons stated- the modern abstraction layers and hardware still (at the central core) run software code from decades before. He said that trying to change over completely to new systems is like trying to change all four tires on a race car while driving over one hundred miles an hour! You have to completely stop to allow the change over to take place.
At 0:45 he points out the OCR characters at the bottom of the check. If the machine can't read them because they are smudged or the check is torn they are routed to another machine that can read them. I interviewed to repair that machine at my first job interview after the Navy. It was called the Lundy--Mae and it was made by Lundy Farringrton it had four hundred and forty circuit cards. Fortunately, I did not get the job because it only paid about $10,500 a year. This was in 1979 but even then I would have struggled in NYC on that income.
"Fortunately, I did not get the job" So why did you apply and waste the interviewer's time? Particularly if you might have got the job whilst not wanting it whereas another candidate was desperate for the job.
@@CasinoWoyale Because I did not know what it paid going into the interview nor did I have any idea what it costs to live in NYC.
i lovvvvve this presentation! thank you for sharing it :D
9:46 those auto-loading tape drives are always fun to watch. Too bad they cut away without showing the tape loading
the nightly batch processing is why banks, at least in Denmark, don't process payments and cash withdrawals between 23.00 and 06.00 on weekdays, or from Friday 23.00 to Monday 06.00. So if you for instance use an ATM to pull out money over night, the bank won't know till the next morning, and that can easily be abused to overdraft your account by far more than the bank allows. For the first 10 years or so with debit cards the bank owned ATMs didn't work at all at night because of that, but since the mid-1990s or so the transactions are still stored even when the bank is closed, but it's not processed till the next morning
Fun fact: The same software and methods are still in use with some banks. At least the hardware changed. But the IT of most banks is an horrible mix of ancient systems.
ICL/Fujitsu/Amdahl equipment?
Why wouldn't each local office instead:
1. manage local universal identification numbers (LUIDs).
2. print out and mail the bank statements themselves?
3. transmit the transactions and new customer CRUDs back to the head office purely for archival purposes?
firstly because the computers were super expensive and required a ton of space (and electricity), so it was unrealistic for every branch to have their own computer. The computer they show in this film would've cost a couple million £ back then and took up most of a floor in a building. Secondly your suggestion would mean that each branch would not be notified if a customer made transactions through a different branch. All the data is sent to the head office specifically so all transactions from all their branches are processed together so each branch can work from identical data. Due to the limitations of how the computers worked back then and the total lack of internet (or secure communication), this was the most efficient way of doing it.
The computer was purely used for processing transactions, nothing else. The local branches didn't have any computers whatsoever and did everything on paper, so if you went into the bank in the morning and withdrew some money, and then came back later to ask about your current balance, they could only tell you the balance from that morning because your withdrawal wouldn't get processed till after the bank closed. Bank branches didn't get their own computer systems till the mid-1980s, and even then it only served to cache the data sent to and received from the head office, but transactions are to this day only processed in one central location. Banks still largely function according to these old principles
@@thesteelrodent1796 Awesome answer. Thank you. Now I'm wondering about a failsafe against fraud: what if, for example, five people withdrew $100 from an account that only had $300? They were at separate branches and made the withdrawals within hours of each other.
@@thesteelrodent1796When I worked at the bank, updated customer account records were received in the morning by microfilm to be read on the microfiche monitor. Also received were unreadable cheques and credits from the clearing house for manual branch input. We had a Burroughs TC500 with paper tape programming and modem. In early 80s we moved to Burroughs Modular Terminals with the luxury of a monitor and 5.25" floppy drives.
....Sorry, should be 8" floppy!
I think my father worked in one of these shops
so.... the , cash transaction takes place, while that cheque waits at that table to get processed later, right? i'd like to withdraw a fantazillion quid pleace... and would you mind me, waiting right here, at that desk?... you don't happen to have a pen, you could lend me a minute there, while you get my cash?
right. You give the girl a cheque and she writes down that she received it. The information from the cheque is sent to the head office and the cheque itself is sent to the issuer's bank so they can process it as well, but because it's a cheque the amount on the cheque will be on hold until the issuing bank transfer the money to your bank. This is why cheques took several days to clear, and only when your bank has processed that they received the money will it be added to your balance. But they won't let you withdraw more than the current balance - or available amount if you have a credit line - because they know what your balance was the day before.
Event sourcing!
Good video, but quite a robotic description of how primitive data bases work. No real explanation of field names, and how it relates to the entire data set.
Its a high level look at the overall operation. There are MANY ways to handle typical banking functions in a batch mode
That time there was cold war.
And then Bitcoin was invented..
This is all pretty backward even for 1980.
this method is still used, just the hardware has changed.
It's actually pretty good.
the only thing that's change since 1980 is that they don't have to use sequential processing, but otherwise banks still work this way