Benchmarking The Lightweight, Musl-Based Alpine Linux

[phoronix]

Phoronix: Benchmarking The Lightweight, Musl-Based Alpine Linux

Following the KaOS and Void Linux benchmarking as part of our next big Linux distribution comparison was firing up Alpine Linux for some benchmarking...

http://www.phoronix.com/scan.php?page=news_item&px=Alpine-Linux-Benchmarking

[staalmannen]

Awesome! Alpine is my favourite distro. I have just not had the guts/motivation to switch to it from Arch for daily use yet

[pal666]

musl busybox distros intended not for xeons, but for 4 megabyte ram routers

[harningt]

Alpine offers the option of sticking w/ just busybox... or going the full mile and installing many of the "standard" non-busybox items as you wish.

[liam]

If be surprised if this isn't significantly slower in many operations as it includes grsecurity/pax.

[caligula]

If be surprised if this isn't significantly slower in many operations as it includes grsecurity/pax.

Sure but replacing the kernel is a no brainer. I use 4.4 with Alpine.

[staalmannen]

Sure but replacing the kernel is a no brainer. I use 4.4 with Alpine.

is there an up-to-date vanilla kernel in the repos? Last time I tried Alpine on my laptop I had some issues with the grsec kernel

[liam]

Sure but replacing the kernel is a no brainer. I use 4.4 with Alpine.

OK. For me, that's one of its biggest selling points as memory is cheap but security isn't. I was only pointing out that benchmarks won't tell you the whole story, and aren't, in themselves, a compelling way to look at Alpine.

[caligula]

OK. For me, that's one of its biggest selling points as memory is cheap but security isn't. I was only pointing out that benchmarks won't tell you the whole story, and aren't, in themselves, a compelling way to look at Alpine.

Fair enough. I didn't think that people use Alpine mainly for security reasons. I use it because the lack of bloat.

[liam]

Fair enough. I didn't think that people use Alpine mainly for security reasons. I use it because the lack of bloat.

Hmmph. I didn't realise that.
That being the case, I'd think they'd just drop the grsec/pax patches since they certainly don't help make the system any trimmer and definitely impose CPU overhead (sometimes quite a bit). Also, user namespaces might be good enough (that's the primary sandbox method that servo uses... they've done a bit of work with seccomp but don't seem to like its usability).