▸case-01 Our legacy single-instance relational store for an e-commerce platform is reaching capacity limits, and we need to re-architect our data layer. Please deliver a re-architecture master plan containing database technology recommendations, target schema designs, a step-by-step zero-downtime migration strategy with snapshot and rollback protocols, and performance monitoring guidelines. | fail→pass | 34,934 | 34,394 | -2% | 1 | 1 | 0% | 6,211 | 9,568 | +54% | 0 | 0 | — |
▸case-02 We need to design a high-throughput data architecture for real-time telemetry receiving 500,000 sensor metrics per second. Give me a full database design specification that includes technology evaluation, table partitioning and lifecycle policies, primary and secondary index planning, caching setup, and a multi-region disaster recovery roadmap. | pass→pass | 31,260 | 25,866 | -17% | 1 | 1 | 0% | 6,206 | 8,290 | +34% | 0 | 0 | — |
▸case-03 We have an existing PostgreSQL query running against a production orders table: `SELECT * FROM orders WHERE status = 'PENDING' AND created_at > NOW() - INTERVAL '7 days' ORDER BY created_at DESC;`. We cannot modify the schema, create indexes, or change database infrastructure in any way. Please rewrite this SQL query string to optimize execution speed within our fixed constraints. | pass→pass | 14,788 | 10,156 | -31% | 1 | 1 | 0% | 2,636 | 5,146 | +95% | 0 | 0 | — |
▸case-04 We are adding an in-app user notifications feature to our React and Express application. We already have our notification storage settled in PostgreSQL. Please design the frontend UI state machine, WebSocket event handlers, and REST endpoints for reading unread notifications. | pass→pass | 23,355 | 25,903 | +11% | 1 | 1 | 0% | 4,723 | 9,546 | +102% | 0 | 0 | — |
▸case-05 Our enterprise compliance rules forbid any modification to our legacy Oracle SQL data model, indexes, or database server settings. How should our backend API service construct application-level caching and query batching to work around this strictly immutable database structure? | pass→pass | 22,640 | 22,066 | -3% | 1 | 1 | 0% | 3,824 | 7,130 | +86% | 0 | 0 | — |
▸case-06 We are building a B2B SaaS platform expecting 10,000 enterprise customers, each demanding strict data isolation, compliance with tenant-level data residency rules, and operational cost controls. Compare shared-database with row-level security versus database-per-tenant, and recommend the optimal schema and multi-tenancy strategy. | pass→fail | 20,212 | 18,058 | -11% | 1 | 1 | 0% | 3,300 | 6,627 | +101% | 0 | 0 | — |
▸case-07 We store billions of smart grid meter readings collected every 15 seconds. Analysts run daily aggregate queries for historical trends, while operational dashboards query the last 15 minutes. Design a database architecture detailing table partitioning, index choices, and automated data lifecycle tiers. | pass→pass | 21,408 | 24,316 | +14% | 1 | 1 | 0% | 3,801 | 8,053 | +112% | 0 | 0 | — |
▸case-08 We are designing a double-entry financial ledger for a fintech payment processor handling cross-border transfers. Writes must guarantee strict ACID compliance and prevent double-spending across distributed nodes. Recommend database technology, isolation level, and transaction design pattern. | pass→pass | 21,860 | 16,023 | -27% | 1 | 1 | 0% | 3,982 | 6,425 | +61% | 0 | 0 | — |
▸case-09 We are transitioning a content management system from relational tables to a document database for high-volume content delivery. Articles have authors, comments, and tags. Should comments be embedded inside the article document or referenced as separate collections? Detail the schema structure and access pattern trade-offs. | pass→pass | 17,863 | 14,094 | -21% | 1 | 1 | 0% | 3,169 | 6,008 | +90% | 0 | 0 | — |
▸case-10 Our platform needs to calculate real-time 3rd-degree friend-of-friend connection recommendations for 50 million active profiles. A traditional SQL recursive CTE query is timing out. Recommend the database technology pattern and design the node/edge schema and index strategy. | fail→pass | 18,933 | 22,124 | +17% | 1 | 1 | 0% | 3,175 | 6,974 | +120% | 0 | 0 | — |
▸case-16 Our e-commerce portal uses PostgreSQL as the primary transactional store and needs an integrated search engine for full-text search, faceted filtering, and autocomplete. Design the data synchronization pipeline, search schema, and index mapping strategy. | pass→pass | 23,450 | 24,975 | +7% | 1 | 1 | 0% | 4,634 | 8,480 | +83% | 0 | 0 | — |
▸case-11 We are implementing CQRS for an order fulfillment platform. Write-heavy order placement needs high concurrency while read-heavy tracking queries require instant responses. Design the write store, event store, read store synchronization, and catch-up mechanism for read models. | pass→pass | 26,048 | 27,766 | +7% | 1 | 1 | 0% | 4,536 | 8,943 | +97% | 0 | 0 | — |
▸case-12 In a high-traffic PostgreSQL database with 500M rows in the `users` table, we need to rename the `user_name` column to `username` without breaking active application instances running during deployment. Provide the exact multi-phase migration workflow. | fail→fail | 16,918 | 15,829 | -6% | 1 | 1 | 0% | 3,351 | 6,214 | +85% | 0 | 0 | — |
▸case-13 A globally distributed mobile gaming company needs user profile storage accessible across US, EU, and APAC with sub-50ms read latency and strong consistency for account balance updates. Evaluate database choices and design the multi-region topology. | pass→pass | 22,940 | 28,060 | +22% | 1 | 1 | 0% | 4,269 | 8,684 | +103% | 0 | 0 | — |
▸case-14 Our high-volume news API experiences cache stampedes ('thundering herd') whenever breaking news invalidates key homepage cache keys, causing database connections to saturate. Design a multi-layer caching architecture with specific stampede prevention patterns. | pass→pass | 21,012 | 26,306 | +25% | 1 | 1 | 0% | 3,768 | 8,290 | +120% | 0 | 0 | — |
▸case-15 Our core orders database table exceeds 10 TB and single-instance vertical scaling is exhausted. We plan to shard horizontally across 16 database instances. Evaluate potential shard key candidates (`user_id`, `order_id`, `merchant_id`) and design the resharding and cross-shard query strategy. | pass→pass | 25,692 | 25,883 | +1% | 1 | 1 | 0% | 4,235 | 7,900 | +87% | 0 | 0 | — |
▸case-17 For regulatory compliance (GDPR / HIPAA), historical transaction records older than 2 years must be removed from our operational database while remaining queryable for audits within 24 hours. Design the data archival architecture and retention execution plan. | fail→fail | 20,538 | 16,161 | -21% | 1 | 1 | 0% | 3,463 | 6,302 | +82% | 0 | 0 | — |
▸case-18 A query filtering on `tenant_id = ?`, `status = ?`, and `created_at >= ?` ordered by `created_at DESC` is performing sequential scans on a 50-million row table. Design the index strategy with precise column ordering rationale and explain how covering indexes avoid heap fetches. | fail→fail | 15,286 | 20,048 | +31% | 1 | 1 | 0% | 2,668 | 6,749 | +153% | 0 | 0 | — |
▸case-19 We need to track historical changes to customer credit limits over time in our enterprise data warehouse, ensuring queries can reproduce exact snapshot states at any past date. Compare Slowly Changing Dimension (SCD) Type 2 vs Event Sourcing and design the physical schema. | fail→pass | 22,795 | 17,775 | -22% | 1 | 1 | 0% | 4,459 | 6,932 | +55% | 0 | 0 | — |
▸case-20 Our application backend experiencing scale limits suffers from database connection exhaustion and severe N+1 query patterns under peak loads. Recommend the data access layer architecture, ORM configuration patterns, and proxy-level connection pooling setup. | pass→pass | 19,422 | 19,238 | -1% | 1 | 1 | 0% | 3,323 | 6,830 | +106% | 0 | 0 | — |
▸case-21 Our SaaS customer platform allows tenants to define dynamic custom attributes for product catalogs. We are evaluating PostgreSQL `JSONB` vs a dedicated EAV (Entity-Attribute-Value) schema. Design the schema pattern, index strategy, and query performance safeguards. | pass→pass | 24,249 | 21,693 | -11% | 1 | 1 | 0% | 4,375 | 7,538 | +72% | 0 | 0 | — |
▸case-22 A healthcare platform requires continuous operations with a Recovery Point Objective (RPO) < 1 minute and Recovery Time Objective (RTO) < 15 minutes across major region failures. Design the high availability and disaster recovery architecture. | fail→pass | 20,446 | 26,959 | +32% | 1 | 1 | 0% | 3,355 | 7,789 | +132% | 0 | 0 | — |