The service levels, which are often set out in a separate service level agreement (SLA) schedule, will cover:

  • the availability and performance standards to which the services are to be provided
  • the remedies available if the service fails to meet the terms of the SLA.

Particular areas of the SLA to look out for include system availability, support and maintenance, and remedies for unscheduled downtime.

The time a hosted service is operating is called uptime. It is usually shown as a percentage. Care should be taken in understanding how this percentage is calculated because it may allow for service outages which means you may not be able to access the services and your data may not be available for certain periods of time. For example, if a provider specifies an outage as being anything of 30 minutes or more, and the service is not functional for 29 minutes, uptime may still be 100%. You should check whether these outages will be announced in advance and whether they will occur outside of your normal operating hours.

The definition of ‘up’ is also important. Your cloud system may be ‘up’ according to your SLA even if a number of features are unresponsive or not functioning properly, provided that core systems can be accessed by the majority of users. Ultimately, your availability figure should mirror the time you actually need to have access to a fully functional system (or, at a minimum, functional in all critical respects).

Ask your provider for evidence of its history of downtime and the measures that have been taken to prevent similar incidents in future. You could also contact other customers of the cloud provider for references.

Given the nature of the cloud service (and certainly public cloud), support and maintenance should be included as part of the standard pricing model, since this will be required to keep the service operational. However, it may be that only basic support is included in your package, with premium support available at an extra cost.

Pay particular attention to helpdesk opening hours, as well as response times and procedures if there are different support packages on offer. The initial helpdesk response may simply log the problem, with a further call back to provide substantive support, and so the definition of what constitutes a response should be linked to the substantive support. It is useful to look for resolution times, as this will allow you to be aware of when your issue should be fixed.

Like most modern IT systems, cloud arrangements depend on internet availability. Also, your IT equipment will need to be of a certain technical specification to access the cloud service. Generally this will be your responsibility to check, but you should ask whether your provider will offer advice on, and support with, checking the necessary equipment and internet connection required for optimum cloud system performance. Your provider may also advise on contingency plans for internet outages.

Your provider should give a clear explanation of the remedies for unscheduled downtime. Key issues are:

  • Will you automatically receive service credits (in other words, a reduction in charges) in the event of failure?
  • If so, are these set at a meaningful level?
  • Is any further compensation available in the event of serious outages?
  • Can you terminate for persistent and/or serious failure to meet the agreed service levels – this will be better than having to rely on “material breach” which can be hard to define in practice.