What is Digg?45 Comments
- spindrift, on 10/11/2007, -0/+15My brain insists on pronouning that word as "Butterface".
- GMorgan, on 10/11/2007, -0/+8Only 2^64 max file size. Not the last file system we'll ever need then :).
Personally I think Ext4 will simply replace Ext3. It's foremost experts are Linux hackers and that will ensure the in kernel implementation will be sound.
Then, assuming Hans Reiser is found innocent, we will argue over whether Reiser5 should replace Ext4 as standard with most of us saying 'you're nuts, Reiser5 is far too likely to eat your files to go in my distro'.
Other than that I'd get in talks with Sun about redistributable patent licenses for ZFS and do the donkey work ourselves. It would be worth it if only to maintain standards. If we can move all major *nix to one FS then it would be a huge boon. - capiCrimm, on 10/11/2007, -1/+9Since I'm not a MS SQL or Oracle expert It's not really my place to speak, but ***** -- this is Digg.
Have you considered that you may have configured Oracle wrong? - theendlessnow, on 10/11/2007, -1/+8Need at least one good enterprise level filesystem. Closest one so far is reiserfs... and well... it has its fair share of problems and may have more moving forward. So apparently, we NEED a few more.
- thecubic, on 10/11/2007, -0/+6ZFS approximates an LVM and physical layer
BTRFS approximates a physical layer
it sounds innovative to have these filesystems do the work, but they can be done elsewhere, like in software (and hardware) RAID, LVM, and multipathing. I see ZFS being good for Solaris because Solaris doesn't do this well currently, but Linux? Those layers work just fine already! - K3ITHK, on 10/11/2007, -3/+8zfs > btrfs
- forkboy, on 10/11/2007, -1/+6Hire someone that knows something about oracle.. cause you've screwed up, or at least, don't know what you are talking about. A "base" oracle install should be less than 1 gig (of data), From memory, if you install all the example schema's it comes upto about 3gb.
Sounds like you have configured some pretty large datafiles, if you make a 15gb datafile, it'll use the 15gb straight away, even though most of it is "empty". - giid, on 10/11/2007, -0/+5I'm far from an Oracle expert, but perhaps Oracle is configured to use a 16 gig table partition (or whatever it's called) and is only using a small portion of that.
- m0j0j0j0, on 10/11/2007, -4/+9They wouldn't need a new file system if you didn't have to commit gigs and gigs of storage for the simplest of databases. One application we were running on MS SQL was using a little over 500MB of space. 'Upgrading' this database to Oracle produced a 16 gig database, before we started adding data. 500 megs to 16 gigs, WTF kind of overhead could possibly warrant that?!??
- dicerandom, on 10/11/2007, -6/+10"The Btrfs disk format is not yet finalized, and it currently does not handle disk full conditions at all. Things are under heavy development, and Btrfs is not suitable for any uses other than benchmarking and review."
How... useful? - aparsons, on 10/11/2007, -0/+4$30k buys you nothing in the oracle world. Try more like $100k for a 2-cpu box.
- emocat, on 10/11/2007, -0/+4Standard Edition One is $4,995/cpu.
- emocat, on 10/11/2007, -0/+3Oh, yeah, that's what Oracle is known for - "instability".
- DarkJesus, on 10/11/2007, -2/+5Looks promising.
- emocat, on 10/11/2007, -0/+3Oh, you mean like OCFS? Been there, done that.
- lonniebiz, on 10/11/2007, -0/+3Unless Oracle contributes a "free and open source file system" to Linux, I have no interest in it.
- aparsons, on 10/11/2007, -1/+4You didn't calculate your tablespace requirements correctly. You should probably read about how oracle stores data before migrating to oracle.
- m0j0j0j0, on 10/11/2007, -0/+3Both valid points, and things I'll have to run by our Oracle dba. I am also not an Oracle expert by any stretch, having read enough papers by David Litchfield and Pete Finnigan on Oracle security to avoid the platform like the plague. But that's just my admittedly biased opinion ;o).
- rbtopp, on 10/11/2007, -2/+5I don't know what kind of drugs your on but i want some.
- infwonder, on 10/11/2007, -0/+2Didn't you see it is GPL ?
- sturdiva, on 10/11/2007, -0/+2Sure, It's called AFS.
- Fordi, on 10/11/2007, -1/+2Why would that be a boon?
- tonaros, on 11/14/2007, -1/+2First thing that came to mind after reading the title: Butras Butras Gali
- emocat, on 10/10/2007, -0/+1Podcast interview with Btrfs creator Chris Mason here: http://feeds.feedburner.com/~r/OTN_TechCasts/~3/137701946/5796492.mp3
- dtiziani, on 10/11/2007, -2/+3It's sure more useful than your comment...
maybe more than mine too, but I really don't care. Keep up the good work, oracle ;) - SpectralSounds, on 10/11/2007, -1/+2Hurray?
- init100, on 10/11/2007, -0/+1Sorry, but that won't work. You still need at least one local filesystem, both for your clients and for your AFS servers.
- PlatSajack, on 10/11/2007, -5/+6fat16 > all
- kusuriya, on 09/07/2008, -0/+1Yes, But tell me what LVM or Raid you have that does snapshots and automatic healing with as little space usage as ZFS? Most Raids you give up a good chunk of space to do the healing, and snapshots are PITA with LVMs for the most part as I recall, and COW is a big plus...
- Feyr, on 10/11/2007, -1/+2does anyone know how this work compare to ocfs2? why would oracle need a third fs (or more?) for linux ?
im still waiting for them to make their database stable, a 30k$ piece of software that crash randomly doesn't exactly inspire confidence - nybble41, on 10/15/2008, -0/+1RAID/LVM in the filesystem has a number of advantages over RAID or LVM at a device level. For one thing, you have much finer control over the method; btrfs will automatically mirror metadata while striping data, for example, unless you specify some other arrangement. Also, when using mirroring the filesystem can rely on checksumming to select the uncorrupted version. Normal two-way RAID mirroring can only detect that an error exists. Space combining and snapshotting (LVM) are a lot more efficient with access to the allocation data. It should also be easier to get the benefits of RAID with varying disk sizes, since the mirroring and/or striping operate in smaller chunks.
- blackoper, on 10/11/2007, -1/+1I'm not a fan of ZFS because of it's slow read/write speed compared to say xfs...
- sneeka2, on 10/11/2007, -1/+1Mine went for BiterFS. Not sure if that's any better.
- majinboy, on 10/11/2007, -1/+1psst, I think you missed the /sarcasm in Plat's post.
- MiserJ, on 10/11/2007, -0/+0no, cause choice is good and it would be boring sorry.
- gravisan, on 11/30/2008, -0/+0Scorchio!
- neutrinodust, on 10/11/2007, -1/+1all for the low low price of $50,000 per proc.
- kusuriya, on 09/07/2008, -0/+0Come back when the ZFS crew figures something out that allows a proper port of ZFS to GPLed systems. AND YES there is a fuse port but at this point in time fuse is mostly an epic pile of fail.
- sishgupta, on 10/11/2007, -6/+6How many filesystems do we need?
I understand that there are various purposes of most filesystems but some are seeming redundant now. - palmer, on 10/11/2007, -3/+2Pretty sure this should be BTRFS, not "Btrfs". WTF is wrong with people that they now can't even spell abbreviations or acronyms correctly?
- tetsura, on 10/11/2007, -2/+1lal, you're description cut short.
- schestowitz, on 10/11/2007, -7/+5For a change, Ellison and the team do something positive for Linux (they also joined OIN a few months back). Hates off for this contribution that makes ZFS look unnecessary...
- inactive, on 10/11/2007, -8/+1god damn there are too many fses in the world, can't we have 1 fs?
The Digg Toolbar for Firefox lets you Digg, submit content, and keep track of Digg even when you're not on the Digg site. Download the official 
© Digg Inc. 2009
— Content created and posted by Digg users is