What is an SLA (Service Level Agreement)?

A service level agreement (SLA) is an outsourcing or vendor contract that outlines the level of service a supplier promises to deliver to the customer. SLAs outline metrics such as uptime, delivery time, response time, and resolution time. An SLA also specifies compensatory actions when requirements are not met, such as additional support, price discounts, or refunds. SLAs are typically agreed upon between a customer and a service provider, although business units within the same company can also make SLAs with each other.

Here are some common types of service level agreements (SLAs).

Customer-level SLA

A customer-level SLA, or customer-based SLA, is an agreement that covers all of the services used by a customer. A customer service level agreement covers specific service details, service availability, responsibilities, escalation procedures, and cancellation terms.

Service-level SLA

A service-level SLA is a contract that details an identical service offered to multiple customers. For example, if a service provider had multiple customers using its virtual help desk, the same service-based SLA would be issued to all customers.

Multi-level SLA

This type of agreement is structured into multiple levels that integrate several conditions into a single system. This approach is suitable for providers with many customers using their product across different price ranges or service levels, or for organizations with several SLA agreements on the same product, each with different terms for different teams. Multi-level SLAs can also span multiple providers. These differing service levels can be built into a multi-level SLA.

There are several common elements you can include in a service level agreement (SLA).

Agreement overview

An agreement overview includes the start and end dates of an SLA, details of the parties involved, and an overview of the services included.

Description of services

A description of services outlines all services provided within an SLA. It details information such as turnaround times, technologies and applications, maintenance schedules, and processes and procedures.

Exclusions

This section describes all exclusions and exemptions that are agreed upon by both parties.

Service level objective

A service level objective (SLO) is an agreement within an SLA about a specific metric target for a service over a period of time, such as response time or uptime. Both parties agree to key service performance metrics backed by data.

Security standards

Both the service provider and the customer use security standards to demonstrate the security measures and protocols in place. This section also optionally includes non-disclosure agreements (NDAs) and anti-poaching agreements.

Disaster recovery process

An SLA will often detail the process of disaster recovery and outline the mechanisms and processes to follow in case of service failure of the vendor. This section also includes information on the restarting process, including restart times and alerts.

Service tracking and reporting

In this section, performance metrics are agreed upon by both parties. Most customers closely track their service performance. A reasonable baseline for this tracking would be before and after using a new service provider.

Penalties

This section clearly states the penalties, financial or otherwise, that either side incurs if they fail to live up to their SLA obligations.

Indemnification

In some contracts, an indemnification clause will cover additional customer costs in legal fees or damages related to breach of the SLAs.

Termination processes

There may come a time when you want to bring your agreement to an end. In addition to requiring a notice period from either party, the SLA also clearly outlines the circumstances that permit termination or expiration.

Review and change processes

You must regularly review your SLA and any key performance indicators (KPIs) that you are using to measure performance. Any large-scale changes in your requirements need to be recorded in the agreement.

Signatures

The agreement to each item contained in the document is reviewed and signed by authorized individuals and pertinent stakeholders from both sides. As long as the agreement is in effect, both parties are bound by it.

There are several common metrics you can expect service level agreements (SLAs) to cover. These enable you to measure your service provider's performance levels as agreed upon in your service level agreement. Here are some common SLA metrics.

Service availability

Service availability is the amount of time that a provider's service is available for use. This is sometimes measured in a time slot. For example, your SLA might specify that a provider's service will be available for a minimum of 99.9% availability for a specific 12-hour window each day. If you have an ecommerce solution that receives orders 24 hours a day, an SLA that guarantees availability of 99.99% for 24 hours a day would be more appropriate, but it will come at an added cost. To meet these performance expectations and avoid service outages, providers may use solutions such as predictive maintenance to ensure all managed services have high uptimes.

Error rates

Customers monitor error rates to measure how often their IT service provider delivers a level of service that falls below customer expectations. For example, customers can set defect rates as a metric to monitor the performance of a virtual service desk. If the number of negative interactions rises above a certain level, this will be flagged. Another example is software testing, where you can set acceptable error rates for coding.

Security

It's critical to measure controllable security metrics, such as antivirus updates and patches, to demonstrate that the vendor takes preventive measures to reduce unintended access. Protecting a critical component or specific service with additional cybersecurity tools, patch management processes, and technical support can improve stability and help providers meet their SLAs.

Response times

This metric sets out the acceptable response time for a monitor. If the maximum response time is two seconds, for example, then the metric measures the average time from multiple locations. As long as the average from all of the sites is below two seconds, it would be considered sufficient.

Business results

You can measure business results directly attributable to your SLA through the use of key performance indicators (KPIs) that both parties agree to. These should be fixed and quantifiable metrics, rather than more subjective service tracking standards. For example, needing to improve customer satisfaction, without quantifying how that is measured and proven, would not be an appropriate metric.

First-call resolution rates

This metric shows how many customers were able to have their issues solved by the service provider during their first contact with a help desk or chatbot.

Abandonment rates

Abandonment rates are appropriate to measure for customer service providers. The abandonment rate shows how many customers ended their communication before getting an answer to their query from the customer service team. A service level agreement would typically include a very low abandonment rate.

There are a number of factors to consider when setting metrics for your service level agreement. They include the following:

• Motivate with metrics so that both parties can work collaboratively and encourage appropriate behavior

• Agree to metrics that are within the control of the service provider

• Choose metrics that can easily be collected and quantified

• Set a proper baseline from which to begin the contract

• Agree on a manageable number of metrics, as too many will make performance difficult to oversee

• Ensure all metrics are very clearly defined

It is important that customers consistently monitor a service provider's performance against the service level agreement (SLA). Customers commonly run weekly status reports that monitor performance. Consistent failures from either side could incur penalties or indicate when it's time to change service providers.

Most service providers make their data and statistics readily available to customers through a customer portal. Here, you can find information about whether you are eligible for remuneration in the event of continued failings.

If the portal used to monitor services is controlled by a third party, they must be involved in the SLA agreement process. This ensures there's no room for error when it comes to the service expected.

There are several penalties that service providers face if they fail to uphold their service level agreement (SLA) responsibilities.

Service credits

Service credits, sometimes known as service-level credits, are deducted from monies owed under a service-level contract. Service credits are deducted when a service provider fails to meet standards set out in the service level agreement. Service providers can regain service credits in the form of earn-backs if agreed upon by both parties in the SLA. Providers are generally eligible for earn-backs if they perform at or above the agreed service levels for a period of time.

Financial penalties

A service provider can incur a financial penalty for failing to meet its service management obligations. Both parties must agree in the SLA upon the failings needed to bring forward a financial penalty and the amounts required.

License term extensions and ongoing support

In the event of failing to meet its obligations, a vendor can be required to extend the terms of the license or continue the contract without further charge.

A service level agreement (SLA) sets out how a customer and service provider work together and what their individual responsibilities are. On the other hand, a key performance indicator (KPI) measures how successfully people or teams are performing against a group of standards. Typically, KPIs are used to measure internal performance standards, while a service level agreement is more suitable to define standards and service expectations in a customer-vendor relationship.

In cloud computing, SLAs define how service delivery operates across distributed solutions. As cloud services are delivered over shared environments and global networks, clearly documented service expectations ensure customers understand availability targets, performance metrics, and the scope of the services provided.

When working with cloud service providers, a customer-based SLA or a customer SLA outlines the key components in the agreement, which normally include uptimes, escalation paths, and any other defined SLA metrics agreed upon to measure performance. Each agreement clearly outlines what the service provider agrees to deliver and how customers can verify service levels.

After provisioning is complete and the customer workload is running in production, cloud service delivery begins. Both parties use internal performance metrics, and in some rare cases, business process metrics, to ensure alignment with broader business objectives.

SLAs in cloud computing clearly define the responsibility boundaries between the customer and the external service provider. For example, infrastructural maintenance tasks and hardware infrastructure network security fall to the cloud provider, while data management and configuration fall to the customer. A shared responsibility model creates transparency and promotes accountability around usage and technical quality standards.

At Amazon Web Services (AWS), we offer service-level agreements (SLA) and publish service-level objectives (SLOs) for all paid and generally available services. We have separate SLAs for each of our multiple services, with more than 180 separate SLAs currently published on the AWS website. For example, the SLA for our popular cloud storage service, Amazon Simple Storage Service (Amazon S3), can be found at Amazon S3 Service Level Agreement.

When choosing an AWS SLA tier, it is important to understand the differing SLAs for each service you use. The SLA should fit with the SLOs you've defined for your application. Consider how these SLOs will impact the downstream SLA you offer your users for applications running on AWS.

Get started with finding the right SLAs on AWS by creating a free AWS account today.