“Lost time is never found again.”
In Performance Based Contracting (PBC) considering availability of a system or service is one of the highest requirements and priorities for both buyer and seller. But what is availability and how do we describe it within our PBC?
Although we could use various definitions of availability such as those from the Reliability engineering field (e.g. US Military Standard 721C) my colleagues and I simply use the following definition as part of our Key Result Areas (KRAs) structure within our PBCs:
“Availability – Providing users with material / services that are in a known state and ready to meet operational preparedness requirements.”
While we have now defined availability creating an availability performance measure is slightly more complicated given our experience shows there are 4 types of availability performance measures. This variety is illustrated below using an example of a system whose availability can be automatically monitored.
Now lets consider each type of availability performance measure described in the diagram above.
Type 1 Continuous Availability – this type of measure simply compares the amount of working (operational) time vs. total time across a certain review period such as a day (24 hours), week or month. This is a standard measure used by mobile telephone companies, Internet Service Providers, cable / satellite TV companies, etc. where the continuous availability of their service is key to their success. A type 1 availability performance measure is typically represented as a percentage.
Type 2 Continuous Unavailability – where the performance level of the availability is very high (e.g. 99.9996%) it may be simpler to define the amount of unavailability allowed. A type 2 availability performance measure is typically expressed as an amount of time (e.g. 2 hours downtime per month).
Type 3 Discrete Availability – unlike a type 1 or 2 availability performance measure a type 3 does not measure availability continuously. Instead it uses at least 1, possibly more, points in time that defines when to measure availability of the system. For example, it could measure availability at the start of each day / shift. Or it may measure every 3 or 6 hours. The typical reason for the variation from a type 1 / 2 measure is that the continuous measurement of availability may not be possible for some systems or the cost of continuously collecting availability data could be too high. A type 3 availability performance measure can be represented either as a percentage or as a state (e.g. pass / fail).
Type 4a Interval Availability / Type 4b Interval Unavailability – while very similar to a type 1 /2 measure the type 4 availability performance measure is only concerned with availability within a specific interval (window) or time. For example, a buyer may only be concerned that the system is available during working hours such as an electronic Point Of Sales (ePOS) system during store opening hours or a flight simulator during working hours. A type 4 availability performance measure can be represented in same way as a type 1 / 2 measure.
By knowing these 4 types of availability performance measures we are half way there to defining our availability performance measure in our PBC. The next step is to better characterise what availability means (e.g. defining working order and service failure vs. service degradation). But more of this in the next part.