Chapter 13 - The file system

This is a critic of UFS (UNIX File System) and derivatives, which are barely used today, apart from NFS. So it's not relevant but interesting, I've always been fascinated by the different take of file systems.

I like the idea that file systems should support both byte level operations and block/record level operations. The author mentions databases as an example, but DBs bypass the file system for other efficiency reason than accessing as record based.
I definitely agree that text files are less efficient when it comes to record based reading.

Storing the type within the file is also something that is discussed a lot. It has its advantages and disadvantages. Here.

The most important part in my opinion is the "exploiting features of modern hard disk". Which is about placing the files on disk in a way such that they will require as less arm movement as possible. This requires a buffering mechanism to be put in place for good disk scheduling to queue writes and reads in the best way. Algorithms like FCFS (First come first served), SSTF (shortest seek time first), SCAN, CSCAN, CLOOK, etc.. See here.

Quote:Will journaling become prevalent in the Unix world at large? Probably not. After all, it’s nonstandard.

Well, it is today.

Chapter 14 - NFS

The chapter is actually a good dive into NFS.
(Nope I didn't know that the initial patent for Ethernet was held by Xerox.)

It reminds me of cool FS I learned about in the OS dinosaur book: WAFL.

Et Cetera

I like that ending about the constant fight about entropy. It's the general repeating theme of the book.
It also describes the whole experience of working in our field, and maybe the human experience too, to constantly repair, maintain, patch. It's 90%, if not more, of all we do. A love for LISP isn't magically going to fix how the world works.
And it's definitely not limited to Unix either.

Overall, I think the book was a bit too negative for my taste and not constructive enough. I'll keep to what I said in my previous reviews because I think it summarizes my overview of the book.

Quote:There's a definite emphasis and contrast on one side with a rosy retrospection, how the past was perfect and solid.
While on the other side, there's also a feeling that Unix is stuck in time and doesn't move on, isn't made for machines of the future.
It's horrible, but there's really nothing better for now.
The best of the worst.
Unix is stuck in the past, we're stuck in the past of the past, and we'd like to jump to a future we don't know but we'll point out all the faults of the present.

Cheers!