It's not really a well-written article, but one very simple example might be LVM = Logical Volume Management. This Linux facility "abstracts away" where, exactly the information is stored. "Logical volumes" are carved out of "storage pools" which might represent several physical drives. Only LVM knows or cares where the data actually is.
Other, more esoteric filing systems are "key based." You specify what you want to find, and the storage system magically delivers it to you. "Massively parallel cloud-based deployments" often use this sort of thing.
In general, these concepts help to uncouple the dependencies that otherwise might exist in a great many subsystems and applications. With LVM, for example, if a drive fills up you add another to the pool. And, if it starts "making ominous clicking noises," you can take it out-of-service without having to shut anything down. The data is simply "moved off," but applications can continue to access it in the meantime because they don't actually know nor care "where" it is.
|