QuestionQ46

VerifiedApplying site reliability engineering practices

You are responsible for a stateless, web-based API that runs on a single Compute Engine instance in the europe-west2-a zone. The Service Level Indicator (SLI) for service availability is currently below the defined Service Level Objective (SLO). A postmortem investigation shows that requests to the API frequently time out, and these timeouts occur because the API receives a high volume of requests and runs out of memory.

You need to improve the service's availability. What should you do?

Explanation

Because the API is stateless, it can be scaled out horizontally rather than scaled up on a single instance. The root cause of the timeouts is that a single instance cannot handle the request volume and exhausts its memory; provisioning additional instances in other zones and placing them behind a load balancer distributes incoming traffic across all instances, reducing per-instance load (and thus memory pressure) while also removing the single-zone point of failure. This directly increases the measured availability toward the SLO. Adjusting the SLO (A) doesn't fix the service, upsizing a single instance (B) doesn't address the single point of failure or true scalability needs, and a failover-only configuration (D) leaves idle capacity that doesn't help distribute the live traffic causing the resource exhaustion — only active load balancing across zones solves both the load and availability problems.

Learn more

Community Discussion

No comments yet. Be the first to start the discussion!