Main Area and Open Discussion > Living Room
A rant on religiousness about OSes
40hz:
Although Solaris is awfully sexy, I have been playing with it and that file system... that file system...
-iphigenie (December 09, 2008, 07:47 AM)
--- End quote ---
OK, I'll bite.
At least 3 times in this thread you've fawned over the file system, and that makes me quite curious...
Are you talking about zfs?
Hoo boy, it sure looks promising -Edvard (December 17, 2008, 07:33 PM)
--- End quote ---
I wouldn't call that "fawning." Possibly an over-the-shoulder smile or two - but definitely not fawning. ;D
If it is ZFS, there's a nice presentation on it here:
http://www.youtube.com/bsdconferences
This discusses the BSD port of ZFS, but I'd suspect it works the much the same as the Solaris implementation.
Looks like the benefits outweigh the detriments, so just out of curiosity I downloaded the OpenSolaris 2008.11 and am going to give it a fly-by and see if it ends up on a spare disk to play with.
-Edvard (December 17, 2008, 07:33 PM)
--- End quote ---
Edvard: Dude!
When you're done playing with it, start a thread on the topic and share your impressions?
---------------
iphigenie: if he does, please be sure to chime in?
I keep hearing all this great stuff about Solaris. I have very little experience with it, but it didn't seem to be so exceptional as to be obvious to a very interested albeit casual observer like myself. I'd welcome any input.
(I'd be tempted to try it myself, but I'm up to my ears in W2k8 Server and FreeBSD 7.1 RC1 right now, so I doubt I'll get a chance to look at Solaris before year's end.)
iphigenie:
I will do -
zfs is bootable, it is the default file system used for 2008.11 - at least I think it is! Must check now, I might have assumed wrongly.
What I really like about it is that it is what I always wanted a filesystem to be:
Truly virtual. We have now reached the point where we have virtual machines created on top of pooled real machines, and the same thing needs to happen with the filesystem.
Dynamic. I dont have the spare hardware to test this but this virtual nature can allocate extra space to mission critical systems, optimise faster/slower drive usage (not yet supporting solid state drive but that is on the way)
Solid. I have just pulled the plug off the machine while disk activity was happening and no repair or anything needed to be done on reboot. That's the limit of my testing
Ease of use. Create a new "partition" or "virtual machine" in seconds, one command, no waiting. Turn some off, shrink some, grow some. Especially with the virtualization this is cool. I currently have no need for it but next time I have an infrastructure project I would have to consider it seriously.
I hear the performance is excellent too, I know smugmug are starting to use it for their mysql databases, with compression on, and were impressed. I also know that wikipedia has just decided to go for it for all their media hosting.
Enough for zfs - i am by far not a filesystem specialist, just a user who has set up and managed web datacentres a few times and to me it seems that zfs (and some of the additional things in solaris) would remove a lot of the pain we had in day to day operations.
As for opensolaris as a whole, i'm ambivalent - the core is a great package but I am both intrigued by some of the features annoyed by any OS which forces so much down my throat - full desktop with tons of apps preinstalled. My normal mode of operation is BSD core, then everything from ports, and only the minimum necessary... I'm from the school that saw anything that is installed and not used as both a waste of resources and a security risk.
The solaris machine is dual boot with FreeBSD7 and I am tempted to revert to freebsd, because i know where to go and what to look for and how to find what is available. On opensolaris I am unsure what is available, unsure how to go about certain things. I am sticking with it mostly because I am curious to see if I can find all that is promised. And an old nostalgic love for Solaris from when I was a student and my first jobs.
But I have too many things to do and haven't sat down for a straight amount of time with a straight goal, which is why I am unsure and uneasy with it still.
Edvard:
So far, I've fired up the Live CD and poked around.
Indeed, ZFS is default, and the desktop is Gnome with a veritable handful of default-installed apps.
I accidentally started the package manager that looks suspiciously like Synaptic, and it filled up my memory and crashed. No problem, really, this was a Live CD and I only have 512G in this machine, although I would have liked a "Cancel" button.
I'll dig up an old drive and slap it in for fun and see what the package manager will offer me, although my first impression of it smells like Ubuntu with a Solaris kernel and Sun Java pre-installed ;)
Solid. I have just pulled the plug off the machine while disk activity was happening and no repair or anything needed to be done on reboot. That's the limit of my testing
--- End quote ---
Nice. Journalling file systems have come a long way in preventing data corruption from power failures/panic resets but it's good to hear first-hand experience on this one.
Ease of use. Create a new "partition" or "virtual machine" in seconds, one command, no waiting. Turn some off, shrink some, grow some. Especially with the virtualization this is cool. I currently have no need for it but next time I have an infrastructure project I would have to consider it seriously.
--- End quote ---
Seriously cool. Ever since I started seriously poking around computers, this one always hung me.
Why can I not just grow a partition as needed? Shrinking I can understand the difficulty, but at least the other way around should be just as you described. One command, done. Good to hear this one swings both ways.
I'll keep you posted, but I'll do it in another post.
f0dder:
A journalling filesystem won't save you from data corruption, imho. It will save you from having a fubar inconsistent filesystem, yes, but not from corrupted files. What happens if you have a power outage halfway through updating some big index file? :)
That said, iirc ZFS supports inexpensive versioning/snapshots, which could be at least part of the solution to avoiding file (as opposed to filesystem metadata) corruption.
Edvard:
As always f0dder, I can depend on you to correct technical details I get muddled about.
Yes, if you pulled the plug during a big operation, the file is going to get borked, no matter what you do.
What impressed me was the claim that ZFS is almost immune to it.
...All operations are copy-on-write transactions, so the on-disk state is always valid. There is no need to fsck(1M) a ZFS file system, ever. Every block is checksummed to prevent silent data corruption, and the data is self-healing in replicated (mirrored or RAID) configurations. If one copy is damaged, ZFS detects it and uses another copy to repair it.
...
ZFS provides unlimited constant-time snapshots and clones. A snapshot is a read-only point-in-time copy of a filesystem, while a clone is a writable copy of a snapshot. Clones provide an extremely space-efficient way to store many copies of mostly-shared data such as workspaces, software installations, and diskless clients.
ZFS backup and restore are powered by snapshots. Any snapshot can generate a full backup, and any pair of snapshots can generate an incremental backup. Incremental backups are so efficient that they can be used for remote replication — e.g. to transmit an incremental update every 10 seconds.
--- End quote ---
Also, check out http://opensolaris.org/os/community/zfs/intro/
And http://uadmin.blogspot.com/2006/05/why-zfs-for-home.html
(a little dated, but it's first-hand experience...)
Looks like a peach to me.
Navigation
[0] Message Index
[#] Next page
[*] Previous page
Go to full version