@Muchemi, just adding to the power outages scenario. The No1 killer of all software systems is power outages, even if you are in a building that has UPS and generators keep in mind that these generators do not run 24hours cycle on a power outage. All generators that are not water-cooled have a certain cool down cycle time. Therefore, between 9pm and 8am, the generators will be off as there is no need to keep the power supplied to the building except external lighting and elevators ( landlord view, passing on all fuel costs to tenants for any additional power requirements ). You must implement a software power-down of server upon exhausted UPS batteries, which will release all the system resources ( whether read/write ). The worst part now is the Database system, if this is shut down without stopping all the processes, you are looking at a worse case scenario. The hard-disk integrity is also affected, so the following morning you can have a system that will not boot or might need serious recovery.

 

And finally to add, ensure that your juniors in the dept know the Unix-clone as well as you. Otherwise, you are going to have to baby-sit the system because no one else knows, so kiss any vacation time goodbye. This is a dangerious pitfall. Your end target is to ensure an uptime of 99% per annum, there's much to learn and much to pass on to others.

 

Rgds. :-)