Why do people separate application servers directly from database systems... even though they have to buy two servers.
What is the rational behind this????
If failure rate of quality components is say 0.27% chance a year per machine (made up figure for home made server... 1 day out of a year pretend for something very active and x86) then you increase your chances for more downtime if you introduce a whole other server into the equation to 2 days out of a year (or whatever expected downtime is...)
I never understood it. Especially if the application requires low CPU usage, but puts through tons of database inserts and selects.
What is the rational behind this????
If failure rate of quality components is say 0.27% chance a year per machine (made up figure for home made server... 1 day out of a year pretend for something very active and x86) then you increase your chances for more downtime if you introduce a whole other server into the equation to 2 days out of a year (or whatever expected downtime is...)
I never understood it. Especially if the application requires low CPU usage, but puts through tons of database inserts and selects.
