You're thinking like an engineer. Engineers are poor at estimations, poor communicators and can't give me a damn budget. I'll never understand why engineers don't see that abstraction, even through corporations is bad in one area, but good in others (libraries, dependencies).
Zapier? Tines? Workato? That's $10B in marketshare that disagrees with this, without even considering the order of magnitude of revenue present in companies like IBM, Oracle, Salesforce, SAP, etc.
And with these tools I know how long it'll take because it's 90% already built. The SLA's and BCP they provide allow me to delegate risk "enough" - the insurance company agrees - just the same as I delegate risk for my authZ/authN to Okta and Auth0, and my infrastructure and pretty much the entirety of my business to one-or-many cloud providers.
Do I wholly agree with the above? No. I'd sooner we all moved to paper-based databases, and basic HTML forms linked up to printers. But I have to be realistic.
The Zapiers of this world do not disagree.. they've built the hub and all the apps are spokes on their wheel. Engineers know that abstractions only help to create good architectures and high-level understanding, but they don't actually make the magic happen on the ground.
> You're thinking like an engineer. Engineers are poor at estimations, poor communicators and can't give me a damn budget.
And yet it is engineers who will be building this magical platform that you think is 'neat', that you think engineers generally somehow can't understand.
How is that going to work out?
Engineers are always greener on the other side of the fence.
Zapier, Tines, Workato (and doubtless some other gibberish names I haven't heard of) don't intend for a system like this to work, either. They won't bend over backwards.
The SSO example is really different from the other ones. We've got a few standards implemented there and the authN/Z interaction is well defined.
For other systems we don't have that. It's a customer in one system, page viewer in another and a spreadsheet row in the billing service. Even with some common format, there's just too many specialisations you need to rely on custom fields.
There's a good reason zapier has an option for "just give us the code to run".
The abstraction is visible. It's also important to remember vendor lock-in, and how they can raise rates and start squeezing out all of your profit margin.
Zapier? Tines? Workato? That's $10B in marketshare that disagrees with this, without even considering the order of magnitude of revenue present in companies like IBM, Oracle, Salesforce, SAP, etc.
And with these tools I know how long it'll take because it's 90% already built. The SLA's and BCP they provide allow me to delegate risk "enough" - the insurance company agrees - just the same as I delegate risk for my authZ/authN to Okta and Auth0, and my infrastructure and pretty much the entirety of my business to one-or-many cloud providers.
Do I wholly agree with the above? No. I'd sooner we all moved to paper-based databases, and basic HTML forms linked up to printers. But I have to be realistic.