Cloud computing is the delivery of on-demand computing resources and services over the internet on a pay-for-use basis. We examine the key components of cloud service level agreements (SLAs), which govern acceptable service delivery and satisfactory customer usage of cloud computing services.

From a technology perspective, one of the key components of the European Commission’s ‘Europe 2020’ vision is the digital economy. In 2014, a Cloud Select Industry Group sub-group published standardised guidelines for cloud computing service level agreements (SLAs) between cloud service providers and business customers. In July 2016, the SLALOM cloud computing initiative – building on this earlier work – delivered the final version of its model contractual terms and SLAs for cloud computing. SLALOM describes itself as a “major step forward for the cloud industry, promising to lower the barriers to entry for SME cloud providers and make it easier for end users to safely migrate to the cloud.”

But if a supplier decides not to adopt the voluntary SLALOM terms immediately, what should a customer look out for when faced with the supplier’s standard SLA? And what are the core provisions that all SLAs should define?

What is the cloud?

Cloud computing essentially involves a model of delivering technology through on-demand network access to a shared pool of computing resources. When a customer purchases cloud computing services, it is purchasing a ‘virtual machine’ that behaves like a physical computer, but which actually utilises resources from numerous interconnected servers. The customer in turn benefits from economies of scale and avails of lower costs in obtaining their IT services. There are different forms of cloud services - a customer may access software (SaaS), infrastructure (IaaS) or platform applications (PaaS). Cloud services can be provided through public networks, private networks or a mix of public/private networks, known as hybrid.

What is a service level?

A service level is a mechanism that defines the expectations and level of service that a cloud service supplier must meet. Service levels are usually set out in the section of a contract known as a service level agreement, or SLA, which is also sometimes known as the cloud service agreement (CSA). Think of the SLA like the ground rules for the cloud service – a robust SLA can protect the stability of the cloud service, protect the customer’s access to data and minimise the expense of any required remedial action. And, if the supplier fails to meet the defined level of service, the customer will usually be entitled to some form of remedy, such as a service credit.

Core criteria

A fit for purpose SLA should contain the following three key criteria:

  1. Availability – uptime
  2. Data access
  3. Performance – response and fix times

While not always workable, a business customer should seek to customise the SLA or to negotiate the inclusion of additional provisions. As the market for cloud services becomes more competitive, it is common to see suppliers being more flexible when it comes to agreeing to bespoke SLAs.

1. Availability – uptime

From a commercial viewpoint, at the core of a cloud computing service contract is the customer’s right to access and use the service ‘on demand’. Given the nature of technology there may, however, be circumstances where the service becomes unavailable, which is sometimes referred to as ‘downtime’. Accordingly, the SLA should directly address the levels of availability or ‘uptime’ that the service provider is proposing to offer.

SLAs usually state the availability of the service to the customer in percentage terms eg 99.5% uptime. The manner in which the service availability percentage is calculated requires close attention from the customer. For example, an availability service level measured over a month compared to one measured over a year can have a significant impact on the calculation of service levels and payment of any service credits.

As cloud services becomes more complex and interconnected, when procuring a bundle of services, a customer may want to ensure that the SLA classifies the non-availability of a single service component as ‘downtime’. In turn, this should assist in assessing whether the guaranteed service levels have been met.

2. Data access

Customers should take care to ensure that the SLA clearly defines what is meant by ‘availability’ and that it covers access to data. For example, if a customer using a cloud storage service can access its data or files previously saved on the cloud but cannot upload new data, it is questionable whether the service is technically ‘available’. Software bugs or accidental deletion can also affect the customer’s ability to access its files. A savvy customer would ensure that this section of the SLA addresses data back-up, data protection and redundancy/disaster recovery.

3. Performance – response and fix times

The customer should consider the performance and support levels that the cloud supplier is offering, in particular, the response and fix times. Response and fix times are often tiered by priority levels, based on the severity of the service downtime, and are usually determined with regard to impact on the customer’s business. For this reason, it is important to ensure that the SLA contains suitable response and fix times so that queries or services most critical to the customer’s business are responded to or repaired with the appropriate priority and are not just aspirational ‘targets’.

Are service credits being offered?

Service credits are the primary financial remedy for a cloud provider’s failure to meet the agreed service levels. An SLA will often include a list of the service credits payable for different levels of service failure. Service credits are often calculated based on a percentage of monthly charges and may be subject to an overall limit or cap.

A customer may try to argue that service credits should be offset against its payment obligations under the contract. The net effect of this is that if a customer suffers downtime, the amount owed on the bill will be reduced.

A supplier will often try to offer service credits as a sole and exclusive remedy for a failure to meet the agreed service levels. This means that the customer should review the level of service credits being offered and assess whether they would represent adequate compensation for the unavailability of the service.

Conclusion - Setting the expectation

The SLA sets the expectation for the relationship between the supplier and the customer. Not all SLAs are the same. Customers should read the fine print and ensure they, or their corporate IT team, undertake suitable due diligence and assessment of the SLA for a new cloud service. An SLA that contains the well-defined core criteria of availability, access to customer data and performance and remediation metrics will, over the term of the contract, save the parties money and improve customer satisfaction.