3 ms·
This is... debatable. I was around in 1985 and if I had proposed using "an eventually consistent schemaless DB" to tackle a bank or a manufacturing project I
by Pamar 8y ago
This is... debatable.
I was around in 1985 and if I had proposed using "an eventually consistent schemaless DB" to tackle a bank or a manufacturing project I would have been laughed out of the room.
And not because it sounded too good to be true, mind you...
- ummonk 8y agoDoes anyone do that, at least on the software side? Obviously there are consistency issues with e.g. non-instantaneous bank teller actions, but those are human inconsistencies, not software inconsistencies..
- Pamar 8y agoWhat I mean is that some of the powerful components we can leverage today solve problems that did not exist in 1985, so having these available then would not really help. I mean: what would you use Node.js for - in 1985 - even assuming you had access to a system to develop and test stuff made with it?
- sedachv 8y ago> I mean: what would you use Node.js for - in 1985 In 1985, your hypothetical bank would be running VMS, which had asynchronous IO system calls as the default. There was no need to "invent" something like Node.js.
- woolvalley 8y agoFunny thing is, that is what ACH and many other banking systems are today, or at least back then. "Reconciliation" can take days.
- sedachv 8y ago> I was around in 1985 and if I had proposed using "an eventually consistent schemaless DB" to tackle a bank or a manufacturing project I would have been laughed out of the room. Medical record systems were running on MUMPS (schemaless) and being eventually consistent (records were keyed in from paper forms) long before 1985.