System Availability and Downtime
The institution must decide precisely when it requires the system to be made available to users. An initial response to this question might be ''24 hours a day for 7 days per week ('24x7') because that is what users would want'', but that ignores the cost implications of extended support and the requirements for system housekeeping and maintenance.
It should be noted that the internal support service model referred to above requires service cover to the organisation for Monday to Friday, 7am to 7pm and to run an unattended system monitoring service outside these core hours, with paging of relevant internal staff to resolve system problems. Clearly, those levels of support come at a substantial cost and it should be determined whether users really do need to access the system outside normal working hours on weekdays and at weekends. This appears to address the cost implications of extended support, but still ignores the requirements for system housekeeping and maintenance.
It will probably be necessary to plan periods of potential downtime for essential system housekeeping and maintenance. These periods may be, for example, between 7am and 9am on each Tuesday and Thursday and could be referred to as vulnerable periods when normal service will be offered unless essential maintenance is necessary. Care should be taken that the 2 hours allowed as vulnerable periods is, in fact, long enough - to avoid disappointing users if downtime is longer. It will be necessary to publish the hours of service for which the system will be available, with the exception of the agreed vulnerable periods, throughout the user community.
Notwithstanding this notification, there will be times when the system will need to be made unavailable for other maintenance purposes such as hardware and/or software inspection and repair. It would be preferable if these activities could take place outside of the normal working day and, again, that sufficient time is allowed for their completion. It is essential that announcements of these service disruptions are widely publicised with a minimum of 24 hours notice. These periods of downtime and the vulnerable periods should be classified as 'Planned Downtime'.
Of course, there will be system failures from time-to-time, but it is hoped that regular system maintenance and back-up and recovery procedures will prevent these happening too often and will enable the system to be made available again as soon as possible. These periods of downtime should be classified as 'Unplanned Downtime'.
The support team should regularly provide statistics of system usage levels and performance, including periods of system availability and planned and unplanned downtime. The performance of the whole system configuration and support function can then be measured and monitored continuously.
