From payroll to payments, e-commerce to cloud hosting, the world of IT service provision continues to evolve. For many of these tasks, businesses adopt cloud-based solutions, but ensuring a robust cloud service with minimum downtime and optimized uptime is challenging.
For every IT service, there is always an expectation: that the services offered meet the needs of those who use the services (and those paying for it). Of course, a given service can be developed without any involvement of customers, but you risk that customers may not want it, as the service may not support their desired outcomes.
Modern organizations consider customers and users as invaluable contributors throughout the IT service lifecycle and desire to collaboratively partner with them in value creation. The shared view of the customer’s expectation and the service provider’s capability is encompassed in a service level agreement (SLA).
In this article, we'll share the ins and outs of SLAs, including templates for creating both simple and more complex SLA documents. We’ll also talk throughout about best practices for protecting the rights of both parties (service provider and customer) and building mutual trust.
SLAs are formally defined in the service management practice of the ITIL 4 framework for IT service management. Let’s start with a definition:
Service Levels Agreements (SLAs) are contracts between service providers and customers (or any two parties) that identify and describe the services required and the expected level of service. SLAs should include:
SLAs are the place to be specific: include the exact uptime or response time in SLAs. This helps customers understand what to expect from your enterprise. It also ensures both stakeholders can hold each other accountable in case of any issues.
(Related reading: SLA vs. SLI vs. SLO: Understanding Service Levels.)
An SLA can have a variety of formats, formality, and customer involvement. You can have:
For example, AWS cloud service plans come with fixed terms and conditions, so customers choose the service levels based on their budget, as prices determine the quality received, including the service performance and support.
Customized services that are based on specific customer requirements will usually incorporate service level targets in the design, and hence the client pays for the level of quality in advance.
Here are some examples of SLAs from organizations that have included some unique features:
The definition and agreement of SLA targets require alignment with internal and external stakeholders who play a role in supporting them. Incorporation of experience metrics measured from the customer perspective helps to prevent the “watermelon” effect for service reporting. When you have this watermelon effect, here’s what happens:
A regular review of the performance against targets and actual workload against the set limits should be held by the service provider and the customer (or business unit representing the customer) to ensure they remain relevant within the evolving operational context.
According to the ISO/IEC 20000-1:2018 standard for service management, the constituent elements that are expected within an SLA based on service requirements are service level targets, exceptions, workload limits, and additional relevant information.
Let’s look at each component:
The service level targets are based on different dimensions such as:
Exceptions apply whenever circumstances arise that would prevent meeting of the targets, such as:
The workload limit measures the applicable volume of work within the SLA, such as:
Other information that may be relevant in an SLA includes:
Now that we understand the background of service level agreements, let’s look at two templates that you can use right now to create your own SLAs.
SLA designs can take a variety of forms, as we’ll see below. What’s most important is to ensure that there is alignment of expectations by all parties involved—as we discussed earlier.
For example: without a comprehensive SLA, very little can be done about poor service when there is no definition of what good service is. Customers will assume that everything will be delivered and available at a 100% level all the time, which is a recipe for disappointment.
A well-articulated SLA formalizes a mutually acceptable position for the service provider and service consumer, ensuring value is co-created and sustained, resulting in a satisfied customer and a fulfilling service experience.
In the following sections, we’re going to show you templates for two approaches for SLAs. One is simple and straightforward, and one is more detailed and may be appropriate for multiple services and/or covering a large contract.
Here is a sample template that an organization providing end-user device support can create to manage SLAs with their corporate customers:
END USER DEVICE SUPPORT SLA
This Service Level Agreement is made and entered into this <Date> day of <Month> <Year> by and between:
<Service Provider Name, Address> (“Provider”)
and
<Customer Name, Address> (“Customer”)
Service Description
The services to be offered are as follows:
Service Level Targets
Service availability:
Service guaranteed performance:
Service support:
Priority | Response Time | Resolution Time |
---|---|---|
Critical | ||
Major | ||
Medium | ||
Minor |
Service Level Exceptions
Exceptional events:
Exceptional time periods:
Exceptional days:
Workload Limits
System Limits:
Human Resource Limits:
Escalation Matrix
Escalation Level | Escalation Timeline | Name & Contact |
---|---|---|
First Line | ||
Second Line | ||
Third Line | ||
Executive Level |
Roles and Responsibilities
Service Provider Roles:
Service Consumer Roles:
SLA Review
A formal performance review of this SLA will be conducted on the first working week of every quarter. Any amendments to the SLA will be discussed and agreed during this review.
Approvals
We have read the above SLA and hereby forge an agreement according to the conditions stated:
Service Provider Representative | Service Consumer Representative |
---|---|
Name: | Name: |
Designation: | Designation: |
Date: | Date: |
Signature: | Signature: |
The elements of a Service Level Agreement (SLA) template may vary depending on the nature of your business.
Since it is the most important document for cloud-based services, you must carefully draft all sections that favor both parties.
In this longer, more detailed SLA template, we’ll outline five sections:
The first page, often called the cover page or title page, is an important introduction to the agreement. Let me outline how you can put up this page in your SLA:
SERVICE LEVEL AGREEMENT
Agreement ID: {SLA-xxx-xxx}
Version: {Add your version number}
Effective Date: {Add your date here}
Service Provider: {Service Provider Name}
Customer: {Customer Name}
Brief Description: This Service Level Agreement (SLA) defines the terms of service between {Service Provider Name} and {Customer Name} for the provision of {brief service description}.
CONFIDENTIAL
Key Contacts:
For {Service Provider Name}: [Name], [Email], [Phone] For {Customer Name}: [Name], [Email], [Phone]
Authorized Signatures | ||
---|---|---|
—-------------------------- | —-------------------------- | |
For {Service Provider Name} | For {Customer Name} |
Any key information goes under this section to provide context and background information about the agreement. This helps both parties understand the purpose of the SLA.
Include a table that specifies parties, their roles, and contact information. To do so, you can follow this format:
Stakeholder | Name | Contact Information |
---|---|---|
Service Provider | XYZ | Email: Phone: |
Customer | ABC | Email: Phone: |
In this sub-section, the service provider must specify conditions for requesting changes. Here’s how you can put it in your SLA:
Add a clear timeline about when and how the contract will end. It could be like:
Include definitions for technical terms and acronyms used in the SLA. Here’s how:
Term | Definition |
---|---|
SLA | Service Level Agreement, a contract that defines terms and conditions between provider and customer |
MTTR | Mean time to repair is the time a system failure recognition and rectification take |
Uptime | The percentage of time that a service is operational and available |
CPE | Customer premises equipment refers to hardware located at the customer's site |
This section includes an in-depth description of the services you will provide. It may include the following sub-categories depending on your needs:
Define the type of service you're going to provide. For example, there are three common cloud services: SaaS, PaaS, and IaaS. So, you must mention which one you provide.
Include your agreement's start and end dates in SLAs to make sure everything is clear about when the terms apply. Next to these dates, add a designated space for signatures and ensure all stakeholders have signed after thoroughly reviewing the agreement.
Here’s how you can put it in your SLA:
Stakeholder | Name | Signatories | Start Date | End Date |
---|---|---|---|---|
Service Provider | xyz | -- | MM/DD/YYYY | MM/DD/YYYY |
Customer | xyz | -- | MM/DD/YYYY | MM/DD/YYYY |
You'll be working with international clients most of the time, so incorporate their time zones into your SLA like this:
Outline the expected performance and quality levels. If your SLA claims to provide 99.99% uptime and 52m 35.7s of downtime, the provider is responsible for ensuring it.
Under deliverables, specify the outcomes you will be delivering. To do so, you can follow this format:
The outcomes that {Service Provider Name} will provide include:
Cloud providers don't care about the quality an individual consumer receives — they rather focus on overall quality. That's why your SLA must have different performance metrics to ensure accountability. The three most important ones to include are:
Let’s see what each of these covers.
Availability defines uptime in percentage — the higher uptime percentage indicates the better quality of the service. Here's what these percentages mean:
Percentage | Downtime per year |
---|---|
99.99999% | 3.2 s |
99.9999% | 31.6 s |
99.999% | 5m 15.6 s |
Unavailability can result in financial loss, so it's not ideal to go below 99.9%. That’s why the service providers should commit to their claimed service availability criteria. For example:
The mean time to repair (MTTR) score measures the outage time. Your SLA must specify the maximum time allowed to acknowledge and resolve incidents and outages. Here's how you can mention it in your SLA:
You must also provide fault-tolerant services. However, outages are part of maintenance and nobody can predict the unplanned outages. That’s why it's better to include the number of planned outages.
For example, mention it like this:
In this section, you define both parties' duties and responsibilities and the consequences of not meeting those service levels. Here are the sub-sections that you must include in your SLA:
You can list the service provider's responsibilities like this:
The {Service Provider Name} responsibilities include:
You can list the customer's responsibilities like this:
The {Customer Name} responsibilities include:
When it comes to SLA violations, both parties include specific clauses and provisions to address potential breaches. The following points are examples of SLA violations:
Here's how to include different services, along with their units and pricing for different tiers:
Service | Unit | Tier | Unit Price | Total Costs |
---|---|---|---|---|
Cloud Storage | 100TB | Tier 1 (first 50TB) | $0.025 per GB/month | $1,250 per month |
Tier 2 (next 50TB) | $0.020 per GB/month | $1,000 per month | ||
Data Transfer | 500TB | Tier 1 (first 300TB) | $0.09 per GB | $27,000 per month |
Tier 2 (next 200TB) | $0.07 per GB | $14,000 per month | ||
API Requests | 5,000,000 requests | Tier 1 (first 2 million requests) | $3 per million requests | $6 per month |
Tier 2 (next 3 million requests) | $2.50 per million requests | $7.50 per month | ||
Total Cost | $43,263.50 per month |
A good SLA protects the rights of both providers and customers. It’s a lengthy process and requires careful planning to design such an SLA. In addition to following our SLA template, here are some best practices that you must follow:
Since every customer has different business needs, one SLA template can’t fit all. The service levels and performance metrics for cloud kitchens will differ from those for a cloud application in healthcare. So, you must communicate with customers and your IT team to create personalized SLAs that set realistic goals.
See an error or have a suggestion? Please let us know by emailing ssg-blogs@splunk.com.
This posting does not necessarily represent Splunk's position, strategies or opinion.