WinBtrfs 1.0 Released For Supporting Btrfs On Windows

[phoronix]

Phoronix: WinBtrfs 1.0 Released For Supporting Btrfs On Windows

Mark Harmstone has released version 1.0 of his custom Windows driver for supporting Linux's Btrfs file-system under Windows 7 and newer...

http://www.phoronix.com/scan.php?page=news_item&px=WinBtrfs-1.0-Released

[sdack]

I wonder how fast it is when compared to NTFS. Are there any tests being planned?

[cl333r]

Awesome, except installing it isn't easy and it's unclear how stable it is. If I wasn't that greedy I'd even donate to that project.

[Michael_S]

sdack
NTFS is pretty slow, but it's not clear to me how much of the slowness is fundamental because it supports a fancy permission system with Access Control Lists (ACLs) and how much is just poor design. Using git on my work computer with NTFS and our medium size code repo is agony, and my SSD is brand new.

[duby229]

sdack
NTFS is pretty slow, but it's not clear to me how much of the slowness is fundamental because it supports a fancy permission system with Access Control Lists (ACLs) and how much is just poor design. Using git on my work computer with NTFS and our medium size code repo is agony, and my SSD is brand new.

NTFS is sorta slow on linux, but it's not that bad on Windows. Windows biggest problem is that a dozen or more processes and services that just won't let the harddrive idle. Windows feels the need to constantly thrash the drive, that it's biggest problem. It's not that NTFS performs badly, it actually performs decently.

EDIT: It boils down to the concept that capacity exists so it should be utilized. Just because you have a terabyte drive means that you should thrash it in every imaginable way until it's full. I know it's a completely retarded concept, but that's how many people feel.

[pb2004]

sdack
NTFS is pretty slow, but it's not clear to me how much of the slowness is fundamental because it supports a fancy permission system with Access Control Lists (ACLs) and how much is just poor design. Using git on my work computer with NTFS and our medium size code repo is agony, and my SSD is brand new.

NTFS isn't slow. Antivirus and compatibility are main factors to perceived slowness. Compare ntfs performance on data volume with disabled and stripped 8dot3 names (on system volume for compatibility short names must be enabled) and disabled antivirus. More info on 8dot3 names performance penalty[1].

1. https://blogs.technet.microsoft.com/...ort-names-too/

[petteyg]

NTFS is exceedingly slow. Especially for large quantities of small files. Try a git clone of KDE's Oxygen icons. Takes many times longer on NTFS than FAT32 (or any Linux filesystem). I was excited when Microsoft first announced ReFS because I thought they might finally be replacing their crappy ancient filesystem, only to see them abandoning their plans, and making it server-only. Compare compilation speed of javac between NTFS and some other filesystem for 75 source files. Plain directory navigation and read access is molasses.

[arokh]

NTFS is much slower on Linux because there's no kernel driver only FUSE which is userspace.

[sdack]

NTFS is much slower on Linux because there's no kernel driver only FUSE which is userspace.

I was asking this question about the performance not because of Linux or the Linux kernel, but when WinBtrfs turns out to be an effective implementation of Btrfs on Windows then it wouldn't be just some gadget utility for accessing Linux disks, but it would also be a serious competitor to NTFS itself. One could then choose to format disks of a Windows PC with Btrfs only for gaining a better performance.

It may still be awkward and perhaps not straight forward to install and to use at this time, but when it performs well, then it's just a matter of time before enough users pick it up and give it a better interface and installer. Just remind yourself that this is an open source project and is being hosted on GitHub.

[PuckPoltergeist]

NTFS is much slower on Linux because there's no kernel driver only FUSE which is userspace.

There is an in-kernel driver for NTFS. It was benched against ntfs3g and it showed ntfs3g was much faster. So FUSE is not an argument. The benchmarks I remember, showed ntfs3g comparable against native Linux filesystems and native Windows NTFS. The main drawback of ntfs3g is the very high CPU utilization.