An enterprise media streaming platform operates a real-time dynamic ad-insertion service hosted on Google Cloud. The executive team reports that service unavailability during high-profile live sporting events results in immediate contract penalty fees and lost ad revenue. However, minor delays in ad retrieval during standard on-demand video playback have negligible business impact and do not trigger financial penalties. The operations team currently monitors a single aggregate SLI of total successful HTTP requests over total requests across all traffic types, using a static rolling 24-hour alert threshold of 99.9%. Consequently, engineers suffer from alert fatigue during off-peak hours while missing critical SLA breaches during major live events. Which architecture and observability strategy should you implement to align technical service levels with business objectives?
- Partition the metrics into context-aware SLIs that differentiate live event ad requests from on-demand requests, establishing a high-availability SLO (e.g., 99.99%) measured over short time windows with multi-window burn-rate alerts for live events, while maintaining a lower availability SLO for on-demand streams.Answer
- BReplace the static alert by setting the global SLI metric as a fixed 99.99% target percentage goal, and configure Cloud Monitoring to trigger automated multi-region failover whenever total request latency exceeds 200 milliseconds.
- CConfigure the HTTP External Application Load Balancer health checks to synchronously perform a live ad-auction query against downstream advertiser endpoints before returning healthy status.
- DReconfigure Compute Engine autoscaling policies to scale strictly based on regional CPU utilization thresholds of 70% across all ad-insertion worker nodes during live events.