Paradoxically, it is preferable, from the client`s point of view, to measure an availability target on the basis of working hours and hours of work rather than on a 24×7 basis. During a monthly measurement period, a 99% availability requirement on the basis of working/working time allows for 1.6 hours of downtime, while the authorized downtime on a 24×7 basis is 7.2 hours. One-day-a-month downtime are unlikely to be acceptable in practical circumstances. Medium measures operating time by checking the response of the HTTP EMbedly API. Every minute (1) a third-party service (currently pingdom.com) attempts to access the embedded API (z.B pro.embed.ly/1/preview?urls=google.com). If the service does not receive a successful HTTP response – that is, a 2XX or 3XX HTTP response code – this will be considered a one-minute downtime. The unavailability of the Embedded service is calculated from the date you did not notify this unavailability by means at firstname.lastname@example.org. We have the right to change this method of measurement within a reasonable time.
Another approach is for the parties to agree to a termination threshold based on the amount of service credits incurred over a period of time, in order to interweav a right of termination in the event of a persistent default. This provides for an additional right of termination in cases where the service provider performs poorly on several occasions. However, in practice, it is not easy to set an appropriate level for the termination threshold and service providers will often attempt to set thresholds for flexibility in the event of a failure.