▸case-01 Our Node.js API server latency spikes to 2 seconds during peak hours. A developer suggests rewriting the entire application in Rust right away to double throughput. Provide an actionable step-by-step performance engineering plan to address this issue. | pass→pass | 16,965 | 16,457 | -3% | 1 | 1 | 0% | 2,840 | 4,940 | +74% | 0 | 0 | — |
▸case-02 We need to run a 50,000 requests per second load test against our live production payment processing environment tonight to simulate Black Friday traffic before an executive sign-off. Outline the execution steps for this load test. | fail→fail | 15,529 | 23,867 | +54% | 1 | 1 | 0% | 2,614 | 4,290 | +64% | 0 | 0 | — |
▸case-03 We plan to swap out our core PostgreSQL relational storage layer with a newly tuned NoSQL database across all production instances simultaneously during a scheduled 5-minute deployment window. Detail the deployment strategy for this optimization. | pass→pass | 17,509 | 13,684 | -22% | 1 | 1 | 0% | 2,883 | 4,335 | +50% | 0 | 0 | — |
▸case-04 We need to add a new 'Export to CSV' button on the user dashboard that formats user records and downloads the file. We do not have any latency thresholds or performance targets specified for this ticket. Provide the implementation plan for this story. | pass→fail | 12,894 | 13,972 | +8% | 1 | 1 | 0% | 2,283 | 4,272 | +87% | 0 | 0 | — |
▸case-05 A user complained on Twitter that 'the app feels slow', but we do not have APM, metrics, server logs, or profiling tools set up in that environment, nor do we have access to any telemetry data. Tell us which exact database query is causing the slowdown. | fail→pass | 9,579 | 10,339 | +8% | 1 | 1 | 0% | 1,637 | 3,898 | +138% | 0 | 0 | — |
▸case-06 Our non-technical board members want a high-level, 2-sentence summary explaining why cloud infrastructure costs increased by 15% last quarter, without technical jargon or performance deep-dives. Draft this response. | pass→pass | 4,599 | 3,808 | -17% | 1 | 1 | 0% | 897 | 2,685 | +199% | 0 | 0 | — |
▸case-07 We are observing broken trace context across our HTTP microservices when HTTP requests pass through an asynchronous RabbitMQ message pipeline, creating disconnected traces in our APM. What is the standard pattern to preserve distributed tracing context across asynchronous queue boundaries? | pass→pass | 14,217 | 14,853 | +4% | 1 | 1 | 0% | 2,614 | 4,791 | +83% | 0 | 0 | — |
▸case-08 Our e-commerce frontend core web vital Largest Contentful Paint (LCP) is 4.2 seconds due to a hero image rendered inside a hero banner component. A teammate suggested putting loading="lazy" on the hero image to speed up rendering. Evaluate this change and propose the correct solution. | pass→pass | 12,361 | 15,872 | +28% | 1 | 1 | 0% | 2,257 | 4,691 | +108% | 0 | 0 | — |
▸case-09 Our Java microservice experiences periodic 3-second latency spikes every 10 minutes. Memory graphs show a classic sawtooth pattern with abrupt drops. A developer suggests doubling the heap size from 8GB to 16GB. Analyze this recommendation. | pass→pass | 14,303 | 17,729 | +24% | 1 | 1 | 0% | 2,599 | 5,045 | +94% | 0 | 0 | — |
▸case-10 When an expensive calculated cache key expires on Redis every hour under heavy load (10k req/sec), hundreds of backend workers simultaneously query the primary SQL database, causing DB CPU to hit 100%. What caching architectural pattern prevents this dogpiling? | pass→pass | 13,839 | 13,141 | -5% | 1 | 1 | 0% | 2,316 | 4,233 | +83% | 0 | 0 | — |
▸case-11 A SELECT query joining orders and line_items takes 12 seconds to run. The developer proposes forcing an index scan hint manually in SQL. What systematic diagnostic step should be executed before adding index hints? | pass→pass | 8,183 | 8,978 | +10% | 1 | 1 | 0% | 1,426 | 3,476 | +144% | 0 | 0 | — |
▸case-12 We want to validate our backend auto-scaling behavior under realistic traffic patterns. A tester proposed sending an immediate spike of 20,000 simultaneous connections in 0 seconds from a single IP address. Critique this design and outline a proper load testing profile. | pass→pass | 16,613 | 21,584 | +30% | 1 | 1 | 0% | 2,706 | 5,565 | +106% | 0 | 0 | — |
▸case-22 A Kafka consumer group is falling behind by 2,000,000 messages during peak events. The team suggests increasing the consumer application thread count to 50 on a single consumer pod subscribed to a 4-partition topic. Analyze this approach. | pass→pass | 13,404 | 16,572 | +24% | 1 | 1 | 0% | 2,243 | 4,892 | +118% | 0 | 0 | — |
▸case-13 We are migrating our web application to HTTP/2. An engineer suggests keeping domain sharding (assets1.example.com, assets2.example.com) to increase parallel asset download sockets. Is domain sharding recommended under HTTP/2? | pass→pass | 10,678 | 10,828 | +1% | 1 | 1 | 0% | 1,724 | 3,882 | +125% | 0 | 0 | — |
▸case-14 We captured a CPU profile for a Golang service experiencing high CPU utilization. The generated flame graph shows 60% of the graph width occupied by json.Unmarshal. A developer suggests adding 4 more CPU cores to the container spec. Analyze this tradeoff. | pass→pass | 14,967 | 14,071 | -6% | 1 | 1 | 0% | 2,600 | 4,343 | +67% | 0 | 0 | — |
▸case-15 Our REST API starts returning HTTP 500 errors with 'Connection pool exhausted' during traffic surges. A developer suggests setting max_connections to 10,000 on the PostgreSQL server. What is the risk and what is the proper backend optimization? | pass→pass | 15,335 | 15,980 | +4% | 1 | 1 | 0% | 2,542 | 4,649 | +83% | 0 | 0 | — |
▸case-16 We want to serve user profile data from Edge CDN with zero origin latency for returning users, while ensuring the background cache is updated asynchronously when data changes. Which HTTP Cache-Control directive combination satisfies this pattern? | pass→pass | 9,107 | 9,930 | +9% | 1 | 1 | 0% | 1,680 | 3,751 | +123% | 0 | 0 | — |
▸case-17 Our GraphQL API triggers 500 SQL queries when fetching a list of 50 posts with their authors, causing 4-second response times. How should this query execution pattern be optimized? | pass→pass | 14,026 | 13,030 | -7% | 1 | 1 | 0% | 2,569 | 4,319 | +68% | 0 | 0 | — |
▸case-18 Our Kubernetes Horizontal Pod Autoscaler (HPA) scales pods based on 80% CPU utilization. However, during flash sales, user response latency breaches our 500ms SLO while CPU remains at 40% due to external I/O waiting. How should the autoscaling trigger be adjusted? | pass→pass | 13,707 | 13,844 | +1% | 1 | 1 | 0% | 2,322 | 4,509 | +94% | 0 | 0 | — |
▸case-19 We are designing a Progressive Web App (PWA) static asset strategy to minimize network roundtrips for repeated page views while serving instantaneous page loads. Which service worker caching strategy should be implemented for static JavaScript and CSS bundles? | pass→pass | 9,808 | 13,689 | +40% | 1 | 1 | 0% | 1,796 | 4,589 | +156% | 0 | 0 | — |
▸case-20 We want to implement continuous production profiling in our Go microservices to catch memory leaks in production. A developer is concerned that profiling will introduce a 30% latency overhead. How should continuous sampling profiling be configured in production? | pass→pass | 14,695 | 15,842 | +8% | 1 | 1 | 0% | 2,689 | 4,756 | +77% | 0 | 0 | — |
▸case-21 Our frontend bundle size grew from 500KB to 3MB over 6 months without anyone noticing until customers complained about slow mobile load times. How do we prevent this in our deployment pipeline? | pass→pass | 14,934 | 19,168 | +28% | 1 | 1 | 0% | 2,535 | 5,285 | +108% | 0 | 0 | — |
▸case-23 A website's First Contentful Paint (FCP) is delayed by 3.5 seconds because the browser downloads three external 500KB CSS stylesheets before rendering any HTML content. What optimization unblocks the initial paint? | pass→pass | 5,852 | 10,981 | +88% | 1 | 1 | 0% | 1,075 | 4,086 | +280% | 0 | 0 | — |
▸case-24 We routed read-heavy API traffic to PostgreSQL read replicas to offload the primary database. However, users complain that immediately after editing their profile, refreshing the page still shows old data. What is causing this and how do we resolve it? | pass→pass | 11,970 | 16,334 | +36% | 1 | 1 | 0% | 2,110 | 4,892 | +132% | 0 | 0 | — |