What is a cloud SLA?
A service level agreement (SLA) is a contract between a service provider and a customer that ensures a minimum level of service is maintained. In cloud computing, SLAs typically cover response times and service availability, and set the benchmark a hosting provider must meet to deliver a reliable service.
SLAs are usually part of a hosting services agreement (HSA) between the provider and the customer, which outlines the responsibilities of both parties during the contract. They set out the service level the provider agrees to meet, in terms of response times and availability, and cover any exclusions, such as scheduled maintenance.
Why do SLAs matter?
An SLA is similar to a warranty for the service provided, and it is a vital consideration when choosing a hosting provider. Downtime is a key concern for all businesses, with the potential to affect your reputation and your bottom line. Our research found that 1/3 of UK consumers would move to a competitor’s website after less than 30 seconds of an outage, even where they had existing brand loyalty.
SLAs have become more important as organisations move critical systems, data and applications to the cloud, since the responsibility for running them passes to a third party. The SLA shows what the provider commits to, and gives you a basis for holding them to it. Even well-run platforms can have outages, so what matters is what the provider has committed to when they happen.
What should a cloud SLA cover?
The most comprehensive cloud SLAs include:
- Cloud server uptime
- Network availability
- Power
- Hardware replacement
- Performance, including maximum response times
- Security of data, including encryption
- Location of data
- Accessibility and portability of data
- Disaster recovery
- Change management processes, such as changes or updates to services
- Incident response times
- Customer support response times
- Service interruptions
Not every provider covers all of these, so check which are included, and which are described only in marketing material and not committed in the contract.
How do uptime percentages translate into downtime?
Uptime guarantees are usually expressed as a percentage, and small differences in the figure mean large differences in permitted downtime over a year:
- 99.9% allows around 8 hours and 46 minutes of downtime
- 99.99% allows around 53 minutes
- 99.999% allows around 5 minutes
Ask what the percentage applies to, such as the cloud server, the network or the whole service, how it is measured and over what period, since a monthly measurement allows less flexibility than an annual one.
What do SLAs not cover?
Every SLA has limits, and it is worth reading them closely:
- Scheduled maintenance, which is usually excluded from uptime calculations
- Events outside the provider’s control, often described as force majeure
- Issues caused by the customer, such as an application fault or a configuration change
- Third-party services that sit outside the provider’s platform
An SLA also covers the infrastructure and services the provider is responsible for. Your own applications, data and access controls remain your responsibility, so an SLA is not a substitute for your own backups and disaster recovery planning.
What happens if an SLA is not met?
If there is an interruption in service and the SLA is not met, you can refer to your contract and request the agreed remedy, which is usually a service credit or a refund. Check how a claim is made and within what time limit, and note that credits are normally capped at a share of your monthly fee. They compensate for the shortfall in service and do not generally cover wider losses, such as lost sales, which is why the commitments themselves matter more than the remedy.
SLAs, SLOs and SLIs
You may see two related terms. A service level indicator (SLI) is the measurement, such as the percentage of time a service was available. A service level objective (SLO) is the internal target a provider aims for. The SLA is the contractual commitment, with consequences if it is missed, so it is the one to focus on when comparing providers.
How do you compare cloud SLAs?
Questions worth asking include:
- What does each SLA cover, and what is excluded?
- How is uptime measured, and over what period?
- What are the response times for urgent issues, and do they apply 24/7/365?
- How quickly is failed hardware replaced, and which hardware does that apply to?
- What remedy applies if a commitment is missed, and how do you claim it?
- How are changes and maintenance handled, and how much notice is given?
- What are the disaster recovery and data portability commitments?
SLAs are living documents that should adapt as you adopt new cloud services. An SLA assessment should be carried out whenever you add a service, to make sure it has the right service levels in place.
What do our SLAs cover?
Our SLAs cover our hosting services, with two key commitments to note. We guarantee hardware replacement within 20 minutes for customers on our HPE Synergy blade server technology, and a response within 20 minutes for urgent priority issues, with the majority of tickets answered within five. Every ticket goes directly to an experienced engineer, who can act on it.
Our Technical Support Engineer Leo explained in our insight on minimising downtime through fast and effective support why hardware replacement time matters, and what a guaranteed time means in practice.
The next steps
SLAs give you a contractual benchmark for the service you receive, and the detail of what they cover, how they measure it and what they exclude tells you more than any single figure. Comparing SLAs carefully is one of the most useful steps when choosing a hosting provider.
If you are comparing providers, speaking with an experienced managed hosting provider can help. Speak to one of our Cloud Consultants about your requirements, and we can arrange an initial consultation to discuss your needs.
