Linux Sysctl Cleaning To Eventually Erase ~64 Bytes Of Bloat Per Array

[JustK]

Go custom kernel!

$ sudo sysctl -a|wc -l
1069
$ echo "1069*64/1024"|bc
66

Next question is: how many VMs do cloud providers serve per machine?

[zexelon]

Go custom kernel!

$ sudo sysctl -a|wc -l
1069
$ echo "1069*64/1024"|bc
66

Next question is: how many VMs do cloud providers serve per machine?

LOL nice a full 66kb to recover

As for VMs per machine it really depends on what "size" the VMs are (ex. how many cores vs how much mem, etc.) I mean if you have a dual epic blade with say dual 128 core epycs and that is hosting single core micro instances you could have up to 256 instances per blade.

Based on 3x data points here lets go really high and say you recover 150kb per instance thats a huge 38Mb of savings... yeah... that same server if you allocate 8Gb of mem to each instance will need 2TB of memory... so there are probably much better savings in other places.

[mos87]

graftin' 👍

[OskarG]

I think it is newsworthy to at least occasionally acknowledge work like this happening in the background to keep optimizing our kernel.

[geearf]

Please stop making fun of cleanup improvements. Even small, when multiplied by millions, they end up huge.
Even if it was 1 Byte per array, that would be welcome.

I am so fed up with bloat in sofwares that I give whatever I get a warm welcome nowadays.

I feel you.
I remember reading a few month back a test of an old Windows on an old computer vs a modern Windows on a modern computer and for the exact same task the old computer/windows pair was so much faster at it! So much for better performance... I tried to find the link but I couldn't sorry.