True, but Paypal is using a really complex system, not in terms of that is **bleep**, but in terms of, that they have a lot apllications, or layers talking together, all is running on seperated servers (I hope), which means, some parts of their whole system can go down, while others may function correctly.
This is both good and bad:
It's good because you make your system, on an overall view, less depending on functionality, which makes the system more stable in the end.
The bad point, is like this - that the payments might be able to be sent / received, but the activity log, transaction ID's etc might be missing, for x amount of time.
So surely, the system can fail, but it didn't fail badly, because when the system was designed, they might have been thinking on this exact problem:
"How do we make a service, that is stable, where we can make the code having only a small amount of dependencies"
If you design this way, your applications will be stable, it might fail in some specific cases, but it's still better than not being able to do anything for a long period 🙂
When designing applications, you should always keep in mind, that they will fail some day.
