While this is sort of a businessy topic, I'm posting it in the Technical forum on the assumption that more people will be capable of contributing.
A theory I heard recently:
It is becoming cliche to say "bandwidth and disk space are cheap. What really matters in shared hosting is RAM & CPU!"
Everyone seems to believe it.
But is it true?
Is it possible that the apparent CPU/RAM bottleneck experienced by growing hosts is actually a misdiagnosis of disk I/O lag? And as their cheap single-disk 1TB 7200RPM SATA 1U servers' load average starts rising to 5, 10, 20, and things are slowing down, and they're exclaiming "we need another server!", what they actually need is more spindles to handle the I/O?
If that's the case, and their modern multi-core CPU is NOT actually stressed, and the solution would have been to deploy multiple, smaller, faster disks in a hardware RAID configuration, does that then imply that disk space IS actually the expensive resource in hosting?
