Which Filesystem to Choose?

Status
Not open for further replies.
ZFS is a very "heavy" file system, at least when used in an array. I used it when I was using FreeNAS for a while, which was BSD based. My dual core system could hardly keep up and it sucked up all 2gb of RAM it had available. For a home file server, it negated the point when I'd have to leave a semi-high powered system running full time so I stripped it to an Ubuntu install with EXT4 on MDADM Raid1 and have lived happily ever after since.

While ZFS may work using FUSE (it's really just a BSD file system from what I understand), I have a hard time accepting it to run reliably, especially when I don't "need" ZFS features and BTRFS is right around the corner. It just seems like using a "wrapper" on something like a file system.. I don't know. Call me old school but it makes my stomach sit a little funny.

Personally, I use EXT4 for everything. Once BTRFS is finalized, I'll give it a go. But for now, I'm quite content.
 
i think zfs is optimized for oracle dbs so possibly it isnt the best for for your ssytems Jayce. I always like to twinker with things tho. Now I'm reading that Red Hats ASM is a excellent performer for oracle dbs. I wonder if this holds true for other workloads too.

ZFS Best Practices Guide - Siwiki
System/Memory/Swap Space
Run ZFS on a system that runs a 64-bit kernel
One GB or more of memory is recommended.
Approximately 64 KB of memory is consumed per mounted ZFS file system. On systems with 1,000s of ZFS file systems, provision 1 GB of extra memory for every 10,000 mounted file systems including snapshots. Be prepared for longer boot times on these systems as well.
Because ZFS caches data in kernel addressable memory, the kernel sizes will likely be larger than with other file systems. You might configure additional disk-based swap areas to account for this difference on systems with limited RAM. You can use the size physical memory size as an upper bound for the extra amount of swap space that might be required. In any case, monitor swap space usage to determine if swapping occurs
 
i think zfs is optimized for oracle dbs so possibly it isnt the best for for your ssytems Jayce. I always like to twinker with things tho. Now I'm reading that Red Hats ASM is a excellent performer for oracle dbs. I wonder if this holds true for other workloads too.

ZFS Best Practices Guide - Siwiki

It wouldn't surprise me if ZFS is optimized for Oracle databases, as ZFS is yet another thing Oracle managed to intenionally **** up.

Because of Oracle, there is a new fork of ZFS, because ZFS was meant to remain open source and be improved on. Instead, once Oracle got a hold of it they announced in the near future they will close the source code. Evidently Oracle has a representative to communicate back and forth with the Illumos project (Illumos is the new open source community project based on Open Solaris, yet another thing Oracle ****ed up) who is also managing open ZFS. The representative's job is to make sure open ZFS and closed ZFS remain at least compatible so it doesn't confuse the masses. Makes sense, right? Clearly someone is thinking with their rear end. MySQL is likely next on the list, as I read last week the original MySQL developer has already forked it since Oracle retains ownership of it now.

Anyway, </done rant>. You're probably right. ZFS might be optimized for Oracle databases, however it has a rather amazing way of maintaining high efficiency with error checking in RAID arrays too. That's why I went after it, since FreeNAS's minimal requirements suggested it could run on computers 20 years old. The curve ball was using UFS with FreeNAS (UFS has far, far less requirements but naturally isn't as feature packed as ZFS), whereas ZFS eats whatever RAM it can get a hold of to maintain that speed it's well known for.

BSD lovers really talk up ZFS, and for good reason. It's a decent file system. But BTRFS is going to kind of nix that gap, as BTRFS will have a lot of the same features while remaining natively compatible with Linux.
 
BSD lovers really talk up ZFS, and for good reason. It's a decent file system. But BTRFS is going to kind of nix that gap, as BTRFS will have a lot of the same features while remaining natively compatible with Linux.

I think that BSD lovers = ZFS lovers because ZFS was (is?) more or less exclusive to them and *Solaris users. Let's hope that BTRFS doesn't get "oracled".

To adapt a phrase; ZFS is the file system of the future, and always will be.
 
I think that BSD lovers = ZFS lovers because ZFS was (is?) more or less exclusive to them and *Solaris users. Let's hope that BTRFS doesn't get "oracled".

To adapt a phrase; ZFS is the file system of the future, and always will be.

Did you mistype, or did I completely misunderstand?

** Edited due to me being completely incorrect.
 
Status
Not open for further replies.
Back
Top Bottom