Google Site Reliability Engineer Interview Questions

I would add backoff, jitter, timeouts, and limits so clients do not keep hammering the failing dependency. I would also think about circuit breakers and whether the system should stop retrying once the error rate crosses a threshold. The reason I choose this approach is that retries can make an outage much worse if every client retries at the same time. I would also mention load shedding or degraded responses to protect the rest of the system. If the interviewer wants more detail, I would describe where retry logic should live.

I would identify the critical dependencies and decide what can be cached, degraded, retried, or isolated. Then I would design timeouts, fallbacks, and circuit breakers so one dependency cannot take down the whole service. The reason I choose this approach is that resilience comes from planning for partial failure, not hoping dependencies always work. I would also think about queueing, bulkheads, and whether the service can continue in a limited mode. If the interviewer wants more detail, I would connect the design to a specific production failure scenario.

I would automate tasks that are repetitive, error-prone, or frequently needed during incidents or maintenance. The reason I choose this approach is that automation reduces manual mistakes and frees engineers to focus on higher-value work. I would also think about how safe the automation is, whether it needs approval gates, and how easy it is to roll back if something goes wrong. If the interviewer wants more detail, I would explain the trade-off between speed and control. In an SRE role, I would always look for opportunities to make the same action safer and easier next time.

I would parse the logs, extract request duration and service or endpoint labels, and aggregate the slowest contributors. The reason I choose this approach is that a simple aggregation can quickly point to the highest-impact bottlenecks. I would also think about grouping by time window, percentiles, and whether outliers should be separated from typical latency. If the interviewer wants more detail, I would mention how I would validate the results against tracing data. The goal is to show that I can turn raw logs into actionable insight.

I would first verify whether the alert is tied to a real user-facing issue or just a noisy internal metric. Then I would tune the threshold, update the signal, or remove the alert if it is not actionable. The reason I choose this approach is that noisy alerts reduce trust and make people slower to respond when a real incident happens. I would also think about whether the metric should be a page, a warning, or just a dashboard item. If the interviewer wants more detail, I would explain how I would test the alert against past incidents.