The Performance of Ubuntu Linux Over the Past 10 Years (phoronix.com) 110
An anonymous reader writes: Tests were carried out at Phoronix of all Ubuntu Long-Term Support releases from the 6.06 "Dapper Drake" release to 16.04 "Xenial Xerus," looking at the long-term performance of (Ubuntu) Linux using a dual-socket AMD Opteron server. Their benchmarks of Ubuntu's LTS releases over 10 years found that the Radeon graphics performance improved substantially, the disk performance was similar while taking into account the switch from EXT3 to EXT4, and that the CPU performance had overall improved for many workloads thanks to the continued evolution of the GCC compiler.
Re:But... how? (Score:5, Funny)
Re: (Score:2)
Less deadweight?
Re: (Score:3)
Re: (Score:2)
And a big reduction in manageability... (Score:3, Insightful)
after forcing systemd on us!
Re: (Score:1)
The nice GUI dialog box asking the sudo password is nice, but yes dropping stderr and always exiting with zero as the exit status does make things harder to manage.
Great dialog unless you are logging in with ssh (have they absorbed that yet?) on a pty. I guess thats not what you do on a modern OS. Retards.
Re: And a big reduction in manageability... (Score:1)
I hate how this is always the answer the systemd guys give. If Red Hat and Debian both can't get the unit files right then that points to a problem with systemd.
Re: (Score:2)
Re: And a big reduction in manageability... (Score:4, Interesting)
Re: (Score:2)
Re: (Score:1)
systemd never ever drops stderr, could you trolls stop spewing that lie sometime?
I manage about 300 servers running Red Hat, and I've seen that problem hundreds of times when a service doesn't start, and there is nothing in the journal. Starting the service by hand usually clearly shows the error. It makes life much more difficult for us.
Re: (Score:2)
Re: And a big reduction in manageability... (Score:5, Informative)
Just to show an example, look what I found in the MediaTomb unit file:
ExecStart=/usr/bin/mediatomb -d -u $MT_USER -g $MT_GROUP -P /run/mediatomb.pid -l $MT_LOGFILE -m $MT_HOME -f $MT_CFGDIR -p $MT_PORT -e $MT_INTERFACE
That "-l" there means that MediaTomb will not log to stdout/stderr/syslog but that it instead logs to it's own logfile in $MT_LOGFILE so no wonder one will never ever find MediaTomb logs in the journal, they are never sent there by the daemon in the first place.
Re: (Score:2)
It doesn't matter what $MT_LOGFILE is set to, that is besides the point that I'm making. With SysVInit you have to tricks like this but with systemd it's no longer necessary since the journal will group together stdout, stderr and syslog into a single log.
But just to prove you wrong, let's look at "man mediatomb":
-l, --logfile
Do not output log messages to stdout, but redirect everything to a
specified file.
So as you can see the whole point of -l is to not use stdout or stderr
Re: (Score:2)
And, exiting with a zero when there's a failure is even worse. We use Puppet scripts to manage about 1,500 virtual machines so it's really annoying that we can't easily detect the failure.
# systemctl start mysqld
Job for mysqld.service failed because the control process exited with error code. See "systemctl status mysqld.service" and "journalctl -xe" for details.
# echo $?
0
# journalctl -u mysqld
-- Logs begin at Mon 2016-01-11 18:19:52 UTC, end at Fri 2016-02-05 21:36:45 UTC. --
The journal is empty!
Re: (Score:2)
Puppet uses exit statuses which are an out of date concept.
Exit statuses are useless now? WTH?
Re: (Score:3)
Let's party like it's 1994!
Re: (Score:2)
The journal is empty because mysql.service does not start mysqld but the ancient mysqld_safe (which is no longer needed with systemd anyway):
ExecStart=/usr/bin/mysqld_safe
And mysqld_safe by default prevents mysqld from syslog, from stdout and stderr. Instead it redirects all these to /var/log/mysql.err. So it's another case where the people who created the systemd unit file didn't know what they where doing, they simply copied how they did it in the SysVInit script.
Regarding the exit code from systemctl, I'
Re: (Score:2)
Re: (Score:2)
It's not always 0, it's only 0 of everything went ok for the things that the "systemctl start" command is concerned which is that it could schedule the service to be activated. This is just the way that the "modern" init launches like systemd, solaris smf, osx launchd and the upcoming new launcher for BSD (which I don't know the name for) works. With these you no longer run a shell script but instead schedule an event to occur and if that event does not occur, then it can wait some time and then try again (
Re: (Score:1)
> service could encounter an error sometime after it forked
True, but SysV init scripts didn't swallow stderr and hide the error like systemd does. That's the problem.
Re: (Score:2)
Re: (Score:2)
In all honesty... (Score:5, Insightful)
Re: In all honesty... (Score:2)
I use a Core 2 Duo (1.86 GHz, 4GB RAM, 128GB SSD) MacBook Air (2010) as a daily driver, updated to the latest 10.11.2 OS X. Works fine with 2 browsers, MySQL
(10GB indexed database), Ruby on Rails, Webrick, Adobe Acrobat Pro... I often have at least 4 apps open at once (right now: Acrobat Pro, Word, Firefox, TextWrangler, and Clean Text). I have a newer Core i5 8GB MBA I leave at home (I commute via metro and the C2D is "expendable"), so I could use a higher performance machine, just haven't felt the need to
Re: (Score:1)
He's running it a VMware guest on Windows.
Re: (Score:2)
I think see your problem - GNOME 3 desktops (Unity, GNOME Shell) require 3D acceleration for rendering, and I am pretty sure the FX 5200 is no longer supported. A non-accelerated DE/WM should do fine for you (and disable GPU acceleration in Firefox)
Re: (Score:2)
I run KDE on Banana Pi and Raspberry Pi and it works fine. Just sayin'.
Re: (Score:3)
I wouldn't call Unity bloated, I'd call it inappropriately feature rich for low-end hardware.
Would be nice if there were a "I'm on a crappy little ARM core" switch in Ubuntu that reconfigures it to a more Raspbian like system hardware requirement.
Re: In all honesty... (Score:2)
If you have a use case that requires a specific FPS, any of the so called "Desktop Environments" are probably way overkill.
A lightweight WM like Openbox, Fluxbox, enlightenment, or even IceWM is probably better suited.
I use IceWM on my custom HTPC. If works for my needs with very minimal configuration.
Re: (Score:2)
Re: (Score:2)
Re: (Score:2)
I loved 10.04! Best version of Ubuntu ever, IMO.
At the time, Ubuntu was just getting better, and better. I could hardly wait for 10.10.
When 10.10 came out, it had that awful Unity desktop. I hated it so bad, I never used Ubuntu again.
And now they are using systemd - barf.
Re: (Score:2)
I don't really care about systemd as long as it works and is fast, but Unity is definitely a failure. It's been 5 years now and I'm still not used to it, and find it a terrible UI compared to Gnome 2.
Not better for Tablets, not better for TVs, and for sure not better for desktop.
The only good thing I like is the windows shortcut to launch an application with the keyboard, but that would be easy to do on Gnome 2 also.
Unfortunately, Gnome 3 was not better, and the Gnome project abandoned Gnome 2, so the
Re: (Score:1)
Re: (Score:2)
Let's put this another way:
Would you expect the Mac OS X software stack to be more efficient now than a similar software stack from 10 years ago?
How about MS Windows?
It's nice to know that a Linux OS hasn't become a bloated mess over a decade of software upgrades.
Re: (Score:2)
Please, stop the denialism.
Windows 10 works smoothly on an Atom N270 or Core 1 Duo. Unity, on the other hand...not so much.
A basic composited desktop with basic animations does not require that much horsepower. Even a measly GMA950 is more than enough. Even VIA C7-M with Chrome9 graphics ran Windows 7 with no hiccups and all 3D eyecandy enabled.
Linux is terribly bloated and unoptimized.
What about measuring reliability? (Score:5, Interesting)
What about measuring reliability? That's one of the most important performance factors of any system of any sort, including Linux installations.
After all, a Linux system that crashes or that does not even boot will offer no reasonable performance of any type!
When I last used Ubuntu, it used its own init system called Upstart. It generally worked well for my needs.
Correct me if I'm wrong, but as I understand it Ubuntu 15.04 was the first to switch to systemd.
Based on my experiences with Debian, systemd was a complete disaster. After doing routine updates I experienced booting problems on several of my computers. After some investigation it turned out that all were due to various problems with systemd.
While desperately looking for solutions to my problems, I found many other people reporting all sorts of problems with systemd. These are the kinds of problems we never experienced with sysvinit or Upstart or other init systems.
It doesn't matter how fast my computer's CPU is, or how fast the disk is, or how fast the graphics are if the computer doesn't even boot far enough to be usable because the init system crapped out.
Re: (Score:2)
I quit using Ubuntu far before systemd (during the Vista days because Vista). Back then, updates would crash it and then make me spend hours getting it back to a functional state.
I went back to Windows with Windows 7. So that's not a systemd thing as much as a Ubuntu thing.
Re: (Score:2)
Re: (Score:2)
It's not supposed to be beta software so reliability is assumed. If you don't get it then the software is not ready for release. It's not high turnover commercial software where something has to be out by the end of the month whether it works or not.
Re: (Score:2)
What about measuring reliability?
In all seriousness: probably because his benchmark programs don't measure reliability. This guy benchmarks stuff. A lot of stuff. He knows how to do it. This time he's benchmarked a bunch of Linux installs. We learn a little, but not so much. Yes, there are some big differences (like the disk performance going down). But often it's not clear why any of that stuff happened. So not very informative, to be honest.
Re: (Score:2)
The most reliable Linux based system I has was based on Linux 2.2.30(?) anyways, 2.2, and Blackbox as a Window Manager. I played Counter Strike on my laptop and had great framerates, I ran prime number generators, cracked passwords, watched fun opengl screensavers, and everything was always smooth as silk and nothing ever crashed no matter how high the load became.
I would go back to it if I could and could have sub-pixel font hinting. My god the fonts were ugly... but perhaps they would look nicer on a 4k s
Re: (Score:1)
Re: (Score:1)
Just shut up and install Gentoo.
Re: (Score:2)
Well, as a desktop user I'm used to the higher reliability of Ubuntu / Mint and debian made some bad moves before already :
- Debian 6 : to protect your freedom, no wifi drivers are included. (Gave away a desktop that it turns out couldn't access the Internet)
- Debian 7 : no official desktop anymore. Well, there's Gnome 3 but why would you put that on an i486 distro?
- Debian 8 : why should I care? Buying a graphics card for getting drivers that play better with a 3D desktop is asking a bit much. But if I com
Re: (Score:2)
My work machine is i7 laptop (4 cores, not "U" variant) with 16GB RAM, no SSD, just HDD. Both Ubuntu 14.04 and Windows 8.1 are surprisingly slow in some standard operations, most annoyingly in logging in, starting Chrome... Also annoying thing with Ubuntu is that if it uses HDD, then everything else is way too slow. If I tar/untar some really big file, and I browse web in parallel, I see that browser is noticeably slower than usual.
I still wonder how we used to do more-or-less the same stuff on machines tha
Re: Not my experience at all. (Score:2)
Back then all the hardware was roughly the same speed. You couldn't saturate your disk bus with a simple tar because your CPU and memory had latencies measured in 100ns-ms, your disk could catch up. Disks are still roughly the same speed as they were 10 years ago. Also, js and html have become bigger and more of it can be found on random websites. I remember a time when you would optimize websites to fit all text, graphics and code under 50-100kb (~1-2s load time). jQuery alone is that size now and we somet
Surprise! (Score:3)
Software performs better after it's had time to mature and be optimized and bugs removed.
Don't tell anyone (Score:5, Insightful)
If Linux fans find out that a distro is in any way successful, they're obligated to split it into a million competing forks and bitch about it endlessly.
Re: Don't tell anyone (Score:5, Interesting)
There may be lots of Linux distros, but they fall into 3 categories:
1) Fedora-derived distros
2) Debian-derived distros
3) Niche distros
We don't really see fragmentation, but rather specialization.
The Fedora- and Debian-based distros see the most use. Even they aren't very different these days, especially the Debian versions that use systemd.
So the fragmentation you're talking about just doesn't exist any more. It's not 1996.
Re: (Score:2)
We don't really see fragmentation, but rather specialization.
And the alternative to specialization is bloatware, so be glad for specialization.
More than Debian and Fedora/Red Hat (Score:1)
So by lineage al
Re: (Score:2)
It might be a bit harsh to call Slackware, Gentoo, SUSE, Arch, etc. "niche", but I do agree that the majority of Linux installations are either from the Debian lineage, or the Redhat (Fedora) lineage.
Re: (Score:2)
...So the fragmentation you're talking about just doesn't exist any more. It's not 1996.
Great, glad we have just one Desktop... well we don't have that. There's kde, gnome, others.
Great, glad we have just one Filesystem... well we don't have that either. btrfs, gpfs, ext(n), etc.
Great, glad we have just one organized set of files... well we don't have that. We have the right way which RedHat mostly does and some really screwed up distros. Sometimes I think they're like - nobody will ever guess the config file is here! Ha!
Great, glad we have just one way to install Linux.... well we don't hav
Re: (Score:1)
Re: (Score:2)
...Or maybe Sandy Bridges.
I think you responded to the wrong story.
Re: (Score:1)
Re:Meaningless stats (Score:5, Insightful)
The only thing that matters is how snappy the GUI is, try measuring framerates of the change from 2D Gnome to 3D Unity. Also compare open source drivers vs proprietary at rendering the GUI. Users don't care about how many bits a hard drive is transferring per second as they will never notice.
Users do care about data rates to/from a hard drive. Ever install a huge game? Ever try to play a movie from disk while uploading photos to picasa? What about backing up data by copying between hard drives?
I can all but guarantee there will be complaints about how long it takes to copy 20GB of crap between drives. Or the fact that the video is stuttering as thousands of photos are being accessed for upload. You'll probably hear "This computer is really slow" when it's actually the hard drive as a bottleneck. Better throughput and smarter accessing/layout aren't things a typical consumer will talk about, but they certainly will appreciate.
Re: (Score:2)
Yes, but... what users are complaining about isn't really how "fair" it is from a CS perspective. What they really want to know is how they can say my video streaming is a lot more important than my bittorrent client and if there's CPU contention or IO contention or network contention just let the video take priority. Because usually somebody with a server has optimized the IO quite well for the use case with 100 streams and they're all equally important. That's usually not the case on the client, some thin
Linux Mint (Score:2, Informative)
Why just Ubuntu? (Score:2)
All Linux is open source, all use the same kernel, all use the gcc compiler.
Why would Ubuntu substantially outperform other Linux distros using the same kernel, compiler, file system, ect? Why would CPU, Radeon graphic, and HDD performance be substantially different?
Re: (Score:1)
Re: (Score:2)