Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Complexity of FreeBSD VFS using ZFS as an example

hybridcluster.com

Post reply on HN
Print view View original

11–17 of 17 posts

Re: Complexity of FreeBSD VFS using ZFS as an example

#11

Post by Scaevolus »

olavgg wrote: post #9
lewq wrote: post #2

Happy to take any questions :-)

I'm also a very happy ZFS user on FreeBSD. I've tried to use it in a lot of crazy ways. At the moment I'm researching if ZFS and a virtual machine running Microsoft SQL Server can improve query performance(with LZ4/GZIP compression) and improve the backup speed. Though I'm wondering about if there are any plans to add XZ compression? Or faster hashing with Siphash24 for data checksums?

LZMA compression (used in the xz format) is much slower than the disk interface and uses a lot of memory, so it would reduce performance.

Re: Complexity of FreeBSD VFS using ZFS as an example

#12

Post by agapon »

jfb wrote: post #7

This was very interesting, thanks. Sorta-OT, I wonder if there's a modern equivalent of Vahalia's Unix Internals: The New Frontiers [1]? [1] http://www.goodreads.com/book/show/337378.UNIX_Internals

Not exactly the same thing but I found the following books quite useful for understanding how a lot of kernle code works: - The Design and Implementation of the FreeBSD Operating System - Solaris Internals: Solaris 10 and OpenSolaris Kernel Architecture

Still, nothing beats reading the source code :-)

Re: Complexity of FreeBSD VFS using ZFS as an example

#13

Post by agapon »

andyjpb wrote: post #5
lewq wrote: post #2

Happy to take any questions :-)

Have you read the VFS Chapter in Beautiful Code?

Thank you for pointing out that book!

It looks like it has a very good description of how C and generated code are used for object-oriented implemetation of VFS. Plain C is not immediately useful for object-oriented programming.

The goal of my blog post series is to describe how VFS works on a higher level.

Re: Complexity of FreeBSD VFS using ZFS as an example

#14

Post by agapon »

kieransimkin wrote: post #4

Interesting to see some technical details of FreeBSD's VFS layer, I've never delved too deeply into it myself but I have looked at Linux's so it's not completely unfamiliar to me.

I haven't had a chance yet to examine Linux VFS in details, so at the moment I can only talk about Solaris and FreeBSD ones.

Re: Complexity of FreeBSD VFS using ZFS as an example

#16

Post by patrickg_zill »

olavgg wrote: post #9
lewq wrote: post #2

Happy to take any questions :-)

I'm also a very happy ZFS user on FreeBSD. I've tried to use it in a lot of crazy ways. At the moment I'm researching if ZFS and a virtual machine running Microsoft SQL Server can improve query performance(with LZ4/GZIP compression) and improve the backup speed. Though I'm wondering about if there are any plans to add XZ compression? Or faster hashing with Siphash24 for data checksums?

Improving the backup speed should be as easy as quiescing the database, taking a ZFS snapshot after you are sure everything is written to disk, then backing up that snapshot.

Re: Complexity of FreeBSD VFS using ZFS as an example

#17

Post by doughj3 »

angersock wrote: post #15

The article mentions illumos--has anybody got any experiences to share with using that instead of BSD or nix?

Illumos is nix. Per wikipedia, "Illumos is a free and open-source Unix operating system. It derives from OpenSolaris, which in turn derives from UNIX System V R4 and Berkeley Software Distribution. "

When Oracle killed off OpenSolaris, Illumos came about.

Post reply on HN