What is a SaaS Service Level Objective (SLO)?

Service​‍​‌‍​‍‌​‍​‌‍​‍‌ Level Objectives (SLOs) for a SaaS platform are a defining line of our expectation from the service (in a given time span, a certain level of performance or availability is targeted). Setting up SLOs in SaaS is mapped to what users encounter, such as system availability, response time, error ratio, or the rate of successful requests. A sample SLO could be 99.99% availability during 30 days or 100ms as the average latency of a user search. These are standards set within the organization and not part of customer agreements, and should be aligned with user experiences.

Service Level Indicator (SLI) is an exemplary quantitative measure. It serves as a gauge of service quality.

Some usual SLIs are:

•   Service availability
•   Tempo de resposta
•   Number of requests processed per unit time
•   Percentage of successful requests

SLIs can present very precise data on service level, while SLOs are made-up goals for carrying out those observations.

Here is a side-by-side comparison between the three concepts:

For a SaaS provider, the role of SLOs is even more central as a large part of their business depends on customer retention, trust, and revenue, which in turn require a high level of and consistent quality of service. They enable SaaS teams to understand the interrelationship between reliability, feature delivery, and operational tasks by “pointing out” areas of service health that are not up to the mark or need improvement.

Error budgets derived from SLOs allow access teams to allocate resources between feature development and system reliability enhancements. Strategic SLO usage may allow SaaS teams to monitor service health and potentially implement countermeasures in response to disruptions, as outlined in Google’s SRE practices.

SLOs are set based on user-experience-related aspects of the service; in other words, they measure how users see service quality. A key factor is that performance metrics of internal resources, such as CPU usage, may not directly reflect the users’ experience.

Some examples are:

•   Webpage loading duration
•   Track the rate of successful transactions

Start by considering the SLIs that quantify the level of service customers perceive, ensuring that the SLOs are related to customer satisfaction.

SLOs define a target reliability level that “balances” user expectations and the team’s capacity to achieve it. Besides being attainable, an SLO should be aligned with customer expectations and business goals. The definition of an SLO is the performance level somewhere between the lowest acceptable and the ideal, not necessarily the best possible.

These three steps make up the process of setting up an SLO:

1. List the critical user journeys that matter most to your users.
2. Pick the most suitable Service Level Indicators (SLIs) for measuring these journeys and set target levels for those SLIs that are achievable in practice.
3. Determine the observation duration (e.g., 30 days, 90 days, or quarterly), work out the error budget, document corrective measures, and establish clear ownership.

Here‍‌‍‍‌‍‌‍‍‌ are some ideas to think about:

1. Check the status of your SLOs by using dashboards, setting up alerts, and tracking your error budget.
2. Review the burn rate to understand the rate at which the error budget is being consumed.
3. Change your SLOs if there are major changes in the way your product behaves, traffic patterns, or customer expectations.