Reliabilityexternal image 633492584387578386-redundancy.jpg
Having a reliable mainframe consist of having a mainframe that has high availablity and easy serviceability. This is achieved through the hardware and software used in the mainframe, everything from redundant systems to the actual coding used in and on the mainframe.

COBOL, never changed, still works fine, very stable programing language.

For a mainframe to have high availability is must be able to handle large amounts of processes as well as communicate with an extensive network (i.e. a bank's communicating with ATM location out there) for most of the year.

If the system is “up” 99.9% of the time,
it will be “down” 43 minutes/month (or over 8 hours/year)

If the system is “up” 99.99% of the time,
it will be “down” 4 minutes/month (or over 50

If the system is “up” 99.999% of the time,
it will be “down” ~1 minute/month (or over ~5

Most companies will be shooting for 99.99% availability for planned and unplanned maintance.
Everything from updates, replacing parts, or any unexpected downtime that may occur. The best
way to acheive high availability is to plan for these to happen. (ie. software failures) So you must
be ready for any of these to happen at anytime meaning having a close monitor on the mainframe,
which will then lead to a more realiable mainframe.

A mainframe using parallel processing will have heighten the availablity. This is due to sharing the
work load amongst several CPCs (Central Processor Complex) for improved response times, meaning
more work is able to get done at once allowing for higher volumes of processes to be handled.

Achieving a high level of easy seviceability in a mainframe is possible through a couple of means,
redundant parts and incremental upgrades. A redundant system has multiple (usually 2) parts that work
on the same process at the same time. Should anything happen to one part in the system, the process
would still go on because of the redundant piece. In the case of incremental upgrades, any updating or
replacing is rolled across the system so that processes are still able to run while the system is being serviced