EXT4 is improving the Linux experience

It’s a universal truth that all PCs are slow. All of them, no matter the hardware, no matter the OS, they’re slow. The faster it goes, the faster we want it to go. It’s never enough.

Now, when it comes to speed, Windows doesn’t offer much. XP is rather speedy but ancient. Vista is new and so slow it couldn’t catch up to a crippled 100-year old turtle on a hot day. 7 promises to address this issue, but it doesn’t exist yet.

Linux is slow too. Take Ubuntu, or OpenSuse, or Mandriva. They’re certainly faster than Vista, but compared to XP, not so much. And don’t get me started on booting, because Windows always has had the advantage there.

That’s why distributions always promise to be faster the next release, or why Red Hat/Fedora are developing Plymouth (which should make booting faster and more attractive).  I must say I never noticed much difference between releases. Real speed umprovements for me came from using Arch, which is compiled for i686 computers, or ditching Gnome/KDE for lighter alternatives like Openbox.

Enter ext4. I timed some actions on the new XFCE install on my test-system, and to my surprise it’s faster than the IceWM install on another partition on the same PC…which shouldn’t be possible. They’re both Arch, and while XFCE is claimed to be lightweight, it’s still a DE and not a WM like IceWM. Yet, XFCE was faster, not by a big margin, but faster nonetheless.
Then, I started to measure boot-times. In a previous article I compared the boot-times of the IceWM install with those of an XP-install on a laptop, so I knew it booted around 52 seconds. That’s the time from pushing the main button to a fully loaded desktop.

The XFCE install booted in 36.5 seconds. That’s a huge difference, and it has nothing to do with XFCE.

In short, the ext4 filesystem made a DE outperform a WM, and that’s something special indeed. I’ve been using Linux for over 5 years now, and never before has a technology appeared which makes that much of a difference, speedwise.

The only downside is that I had installed and configured my main system just before the latest Arch Linux ISO was released…which means it’s still ext3. And I really don’t feel like going through all that configuration again.

San

Advertisements

15 Responses to EXT4 is improving the Linux experience

  1. You can convert your ext3 filesystems to ext4 just by enabling ext4’s features!

    Just google for something like “convert ext3 ext4” and you’ll find your answers.

    Cheers!

  2. Ron says:

    Look at ext4 on the Arch Wiki, you can fully convert (get full benefits unlike Cassiano method) from a LiveCD (only requirment is that it has, e2fsprogs 1.41 or later), of course attempt at your own risk, Im planning to upgrade my / partition soon via that method

  3. […] EXT4 is improving the Linux experience In short, the ext4 filesystem made a DE outperform a WM, and that’s something special indeed. I’ve been using Linux for over 5 years now, and never before has a technology appeared which makes that much of a difference, speedwise. […]

  4. Raj says:

    >>Linux is slow too. Take Ubuntu, or OpenSuse, or >>Mandriva. They’re certainly faster than Vista, but >>compared to XP, not so much. And don’t get me started >>on booting, because Windows always has had the >>advantage there.

    Eh? What are you talking about? My Ubuntu 8.10 boots MUCH faster than XP with all software installed. If you are talking of a fresh XP installed with no software, then thats a useless comparison.

  5. celettu says:

    Ron: I’ll look into that. Thanks.

    Raj: What does installed software have to do with boot times?

  6. David Gerard says:

    @celettu – a lot of software puts itself into startup to preload, e.g. MS Office.

  7. celettu says:

    David:

    I know, but then I’d measure the boot time of XP + MS Office + whatever is installed.

  8. @celettu: Almost every piece of software that’s installed in Windows puts extra bloat in the registry. Registry bloat and fragmentation is one of the major aspects of XP’s slowness, including boot time.

    I agree with David Gerard that boot time should be measured in something like a “Full XP Desktop (including all used apps)” to a “Full GNU/Linux Desktop (which usually is not much more than a default install)”.

    In this kind of comparison, GNU/Linux kicks XP’s ass in my experience (although others’ mileage may vary).

    @Ron: what exactly is the difference? AFAIK, ext3 is forward compatible with ext4, although ext4 is not backwards compatible with 3 exactly because these extensions that make up ext4 break the compatibility. Would you mind ellaborating?

    Cheers!

  9. celettu says:

    Cassiono:

    I agree that out of the bow, XP offers next to nothing in terms of userful software. I don’t think it’s a very good OS either.

    BUT…when I want to measure boot times and nothing but boot times, I’m going with the defaults.

  10. Zeist says:

    In my mind a “full GNU/Linux desktop” does not include any major heavy bloated pieces of software such as an office suite, a DE or a graphical login manager. This is why Linux will always boot faster than Windows.

  11. Raj says:

    Like Cessano said, it makes more sense to measure boot time to a *usable* desktop. If you just want to measure the boot time of the OS, then for Linux, you should probably remove the time required for the GNOME/KDE desktop since that’s just another app.

    I’m NOT saying that it’s all Microsoft’s fault or something! It’s just that for me, the time it takes to do something useful, such as start Firefox – is much lesser on Ubuntu.

  12. […] using Arch and Openbox, and I love it. But the filesystem is ext3, and as I have said before, ext4 offers a real performance boost. Also, I haven’t installed the 64-bit version, and since Arch’s latest April […]

  13. […] and desktops galore, and never had a problem. They only thing that’s very noticable is the speed boost. Moving […]

  14. Yonah says:

    You made the mistake of refering to Linux as somehow being less than perfect. Notice the result from people like Raj and Cassiano. These reactions represent a real problem for the Linux community, but little has been done about them.

    I’ll make one technical complaint that wasn’t addressed. Registry bloat and fragmentation is a red herring. In fact, while some swear by “registry cleaners” and “defrag utilities”, controlled experiments can show absolutely NO measurable impact on system performance after using such tools. What they do offer is the same psychological benefit of deleting all your browser cookies. You just “think” it’s doing something positive.

  15. @Yonah: You have probably misinterpreted my comment, mate. I have never said that Linux (or GNU/Linux for this matter) is perfect, not even close to it.

    What I said is that boot times (in both Windows AND Linux) tends to take longer the more applications you have installed. Of course that, in the case of a GNU/Linux system this is only true if the application puts a service to start at boot time, whereas in a Windows environment the bloat that every application puts in the registry makes your boot time worse.

    If you really want to measure bare bones boot time, you should probably try a Debian GNU/Linux basic install, without all the optional tasks/packages. THEN you would probably have an actual default-to-default install boot times comparison. And I can bet that Debian would boot a lot faster!

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: