Clear Linux Container Performance Continues Showing Sizable Gains

[phoronix]

Phoronix: Clear Linux Container Performance Continues Showing Sizable Gains

For those interested in using Distrobox to augment your operating system's package selection, not all containers are created equally. Distrobox developer Luca Di Maio recently did some tests for looking at the Linux container performance...

https://www.phoronix.com/scan.php?page=news_item&px=Clear-Containers-2022

[caligula]

On top of that, many containers are full of bloated crap. For example a slim version of container might be fully adequate for the task, but a lazy ass DevOps engineer might not been willing to spend 15 minutes testing different alternatives, when the most bloated distribution already does the job. Only larger FAANG companies care about performance, smaller shops just throw shit at wall, and once it sticks, call it a day.

[arglebargle]

This shouldn't be surprising given that Clear builds all of their packages with -mtune skylake. I'd really like to see some comparison with, say, an Arch x86-64-v3 container using the ALHP repo as it's base. I suspect you'll see performance on par with Clear when using another similarly optimized distro.

The last time I looked into this I found that you can expect a ~15% improvement when compiling the system for -v3 on recent hardware.

[zxy_thf]

Can't see the point of these benchmarks.
They only test Clear in Clear, Fedora in Fedora and native Fedora.

Only additional conclusion is negligible performance penalties to containerize existing Fedora applications,
The performance advantage of Clear is well known and these numbers about Clear in Clear ain't something new.

The meaningful test is to run application inside Clear container with non-clear native OS.
Switching to Clear Linux as the host OS is too costly for most people who use Linux for their daily works.

[89luca89]

Can't see the point of these benchmarks.
They only test Clear in Clear, Fedora in Fedora and native Fedora.

Only additional conclusion is negligible performance penalties to containerize existing Fedora applications,
The performance advantage of Clear is well known and these numbers about Clear in Clear ain't something new.

The meaningful test is to run application inside Clear container with non-clear native OS.
Switching to Clear Linux as the host OS is too costly for most people who use Linux for their daily works.

it is all on the same Fedora host
Clearlinux is only in container
and the Fedora container is used as reference for podman overhead

[89luca89]

This shouldn't be surprising given that Clear builds all of their packages with -mtune skylake. I'd really like to see some comparison with, say, an Arch x86-64-v3 container using the ALHP repo as it's base. I suspect you'll see performance on par with Clear when using another similarly optimized distro.

The last time I looked into this I found that you can expect a ~15% improvement when compiling the system for -v3 on recent hardware.

very interesting! I'll surely try that in the future

[edwaleni]

I tried to run Clear Linux in LXC via Proxmox. No templates exist that I can find.

It was suggested to run Debian in a VM and then run Docker images of Clear there. Kind of defeats the purpose with so many layers.

[89luca89]

I tried to run Clear Linux in LXC via Proxmox. No templates exist that I can find.

It was suggested to run Debian in a VM and then run Docker images of Clear there. Kind of defeats the purpose with so many layers.

You could try with Distrobox
It uses the official image from docker hub, and as showed it does not waste the advantages of Clear