Platform observability and SLAs in MLOps - Practice Problems & Coding Challenges

Challenge - 5 Problems

🧠 Conceptual

intermediate

Understanding Service Level Agreements (SLAs)

Which statement best describes the purpose of an SLA in platform observability?

A An SLA defines the expected uptime and performance targets agreed upon between service provider and user.

B An SLA is a tool used to monitor the internal logs of a platform without user involvement.

C An SLA is a software that automatically fixes bugs in the platform code.

D An SLA is a backup system that stores data in case of platform failure.

💡 Hint

Think about what agreements between users and providers usually include.

SLAs set clear expectations for uptime and performance, helping both parties understand service quality.

💻 Command Output

intermediate

Interpreting Observability Metrics Output

Given the following Prometheus query output for error rate over 5 minutes, what is the error rate percentage?

error_rate{service="ml-model"} 0.02

A 2%

B 0.2%

C 20%

D 0.02%

💡 Hint

Remember to convert decimal to percentage by multiplying by 100.

The value 0.02 means 2% error rate because 0.02 * 100 = 2%.

🔀 Workflow

advanced

Setting Up Alerting for SLA Breach

Which sequence of steps correctly sets up an alert in a monitoring system when the platform's uptime falls below 99.9%?

A 1,3,2,4

B 2,1,3,4

C 1,2,3,4

D 3,1,2,4

💡 Hint

Think about logical order: metric first, then alert, then notification, then test.

You must first define what to monitor, then create alert rules, set notifications, and finally test.

❓ Troubleshoot

advanced

Diagnosing Missing Observability Data

You notice that your observability dashboard shows no data for the last hour. Which is the most likely cause?

A The platform uptime is 100%, so no data is needed.

B The SLA agreement expired and disabled monitoring.

C The notification system is down, so data is not displayed.

D The data collection agent crashed or stopped running.

💡 Hint

Consider what collects and sends data to the dashboard.

If the data collection agent stops, no new data reaches the dashboard, causing missing data.

✅ Best Practice

expert

Choosing Metrics for SLA Compliance Monitoring

Which metric is the best choice to monitor to ensure compliance with a 99.9% uptime SLA for an ML platform?

A Average response time of the ML model predictions.

B Percentage of successful requests over total requests in a time window.

C CPU usage percentage of the ML model server.

D Total number of requests processed per minute.

💡 Hint

Think about what directly reflects availability and success rate.

Percentage of successful requests directly measures uptime and availability, matching SLA goals.