[nzlug] CentOS and the "joys" of rpm based systems

[Daniel Lawson]

>
>>>  For example, why wouldn't you reconfigure the kernel of a database server to reflect its needs, rather than installing one that runs X pretty well?
>>>   
>>>       
>> Mostly because the kernel developers don't think I should have to have 
>> different kernels. Which tweaks did you have in mind?
>>     
> Well, changing the scheduler is a good start.
>   

If you mean the I/O schedular, this is easily done at run-time on a 
per-block device basis (and can therefore be done in an init script), or 
can be done system-wide at boot time.

The kernel developers are trying pretty hard to come up with 
best-effort, all-round CPU schedulers too. There's a bit of flux at the 
moment (perhaps a lot?), but they are aiming for one scheduler to rule 
them all, with run-time tweaks if needed.

I'd concede that if you want a newer IO or CPU scheduler than is in your 
distro's kernel, compiling kernels might be a good option. But then 
again, I'm not sure I'd jump to adopt a brand-new IO scheduler on a 
machine that's busy enough to care about IO schedulers :)