Sometimes you need that extra stuff for your application. Obviously I don’t know what exactly the author is running but I have some experience with sysadmining applications on embedded devices.
You don’t always have the ability to say “we can’t do this because of x y and z reasons” because that you’re calling out is the primary revenue source and no one wants to take the time to improve and optimize it.
I hate that fact very much but that’s the reality of business and living in a scarcity society. So you make other decisions and free up anything else you can.
You also point out 4000Mb,the author states and I agree, it’s not uncommon to see 2000Mb or even 1000Mb. That 3% starts to become more problematic.
If the design cuts it so close that it can’t spare 1% of RAM, the system is in danger of crashing in production due to memory exhaustion.
Sure, at some point the impact of systems will matter and you should choose something else. At the point, you may also want to reconsider traditional Linux altogether in favor of a special low-memory variant or alternative.
Author complains systemd using 0.3% of memory is too much.
Systemd uses more memory than predecessors while current embedded systems often have more memory than predecessors.
Still, I upvoted the post because options and diversity in the Linux ecosystem are healthy.
Author complains about that in resource constrained environments like low memory raspis and other embedded Linux devices.
It’s actually a very valid point where you have 4 or fewer gigabytes of ram.
It’s using 8 of 4,000 MB of RAM.
For a lot of projects, it’s not worth optimizing the thing that’s using 0.3% of the memory.
Surely whatever is using the other 99% is a larger concern.
Sometimes you need that extra stuff for your application. Obviously I don’t know what exactly the author is running but I have some experience with sysadmining applications on embedded devices.
You don’t always have the ability to say “we can’t do this because of x y and z reasons” because that you’re calling out is the primary revenue source and no one wants to take the time to improve and optimize it.
I hate that fact very much but that’s the reality of business and living in a scarcity society. So you make other decisions and free up anything else you can.
You also point out 4000Mb,the author states and I agree, it’s not uncommon to see 2000Mb or even 1000Mb. That 3% starts to become more problematic.
It was not even 3%, it was a tenth of that: 0.3%
Even with 1GB of RAM, it would be 1.2%.
If the design cuts it so close that it can’t spare 1% of RAM, the system is in danger of crashing in production due to memory exhaustion.
Sure, at some point the impact of systems will matter and you should choose something else. At the point, you may also want to reconsider traditional Linux altogether in favor of a special low-memory variant or alternative.