▸case-01 We are designing a cloud-native order processing service on Amazon DynamoDB. The engineering team is debating whether to draft entity-relationship diagrams (ERDs) and normalized tables first, or list application access patterns first. Provide a recommendation on how to approach data modeling for this system. | pass→pass | 17,002 | 17,557 | +3% | 1 | 1 | 0% | 2,984 | 3,493 | +17% | 0 | 0 | — |
▸case-02 An IoT fleet tracking platform receives 200,000 updates per second on Apache Cassandra. Developers proposed using the string 'vehicle_telemetry' as the single partition key for all incoming records so all data is in one place. What issue will this cause and what partition key strategy should be used instead? | pass→pass | 13,706 | 13,851 | +1% | 1 | 1 | 0% | 2,386 | 2,591 | +9% | 0 | 0 | — |
▸case-03 In an Apache Cassandra table storing user chat messages, application queries must retrieve the 50 most recent messages for a specific channel ID ordered by timestamp descending. A developer wants to query all messages for the channel and sort them in application code. How should the CQL table definition handle this requirement at the database level? | pass→pass | 13,463 | 10,878 | -19% | 1 | 1 | 0% | 2,412 | 2,610 | +8% | 0 | 0 | — |
▸case-04 An e-commerce API built on Amazon DynamoDB needs to retrieve a customer profile, their recent orders, and the items inside those orders. The team is considering creating three separate tables (Customers, Orders, OrderItems) and running three sequential HTTP API calls per web request. What design pattern consolidates these entities into a single DynamoDB table? | pass→pass | 8,216 | 14,649 | +78% | 1 | 1 | 0% | 1,560 | 3,212 | +106% | 0 | 0 | — |
▸case-05 In a high-throughput Apache Cassandra database, user display names are repeated across millions of activity post rows to optimize read speeds. A team member proposes removing the display name from post rows and doing runtime JOIN queries across tables to ensure data normalization. How should data modeling handle this trade-off in distributed wide-column stores? | pass→pass | 17,795 | 15,509 | -13% | 1 | 1 | 0% | 2,789 | 3,151 | +13% | 0 | 0 | — |
▸case-06 A multi-region Apache Cassandra cluster with replication factor 3 in each region needs strong read-after-write consistency within the local datacenter without suffering high cross-region WAN latency penalties on every request. Which read and write consistency levels should be configured in CQL drivers? | pass→pass | 8,152 | 10,681 | +31% | 1 | 1 | 0% | 1,576 | 2,458 | +56% | 0 | 0 | — |
▸case-07 An Amazon DynamoDB table tracking user tasks needs a secondary access pattern to query tasks by status (e.g., 'PENDING') across all users. Developers are deciding between a Local Secondary Index (LSI) and a Global Secondary Index (GSI). The table is expected to store hundreds of gigabytes per partition key over time. Which secondary index type must be chosen and why? | pass→pass | 7,526 | 10,017 | +33% | 1 | 1 | 0% | 1,393 | 2,409 | +73% | 0 | 0 | — |
▸case-08 A DynamoDB table has 40 attributes per item. A Global Secondary Index (GSI) is being created to support a high-volume search query that only requires 2 attributes (`status` and `updated_at`). Developers plan to project all 40 attributes into the GSI to support possible future queries. How should the index projection be configured to minimize storage and write capacity costs? | pass→pass | 8,786 | 9,492 | +8% | 1 | 1 | 0% | 1,627 | 2,301 | +41% | 0 | 0 | — |
▸case-09 A Cassandra CQL developer is writing a query to select user transactions between two timestamps across the entire cluster using `SELECT * FROM transactions WHERE created_at >= '2023-01-01' ALLOW FILTERING;`. The query is timing out in production. How should the query and schema be restructured to query range data effectively? | pass→pass | 14,467 | 15,248 | +5% | 1 | 1 | 0% | 2,679 | 3,471 | +30% | 0 | 0 | — |
▸case-10 During a national online voting event, an Amazon DynamoDB table receives 50,000 writes per second for a single item representing a popular candidate's vote tally (`PK: CANDIDATE#1`). The table is experiencing severe ProvisionedThroughputExceededException throttling. How should the primary key strategy be modified to handle this write traffic? | pass→pass | 13,274 | 11,671 | -12% | 1 | 1 | 0% | 2,443 | 2,889 | +18% | 0 | 0 | — |
▸case-11 An application records user clickstream events in Apache Cassandra using `user_id` as the partition key. Heavy users generate tens of millions of events, causing individual partition sizes to exceed 100 MB and triggering tombstone and JVM garbage collection issues. How should the partition key design be revised? | pass→pass | 16,147 | 15,561 | -4% | 1 | 1 | 0% | 2,815 | 3,306 | +17% | 0 | 0 | — |
▸case-12 In an Amazon DynamoDB single-table design, students can enroll in multiple courses, and courses contain multiple students. The application needs to query all courses for a given student, and all students for a given course. How should the table keys and Global Secondary Index (GSI) be configured? | pass→pass | 12,994 | 10,628 | -18% | 1 | 1 | 0% | 2,756 | 2,805 | +2% | 0 | 0 | — |
▸case-13 A Cassandra CQL microservice soft-deletes individual columns by issuing thousands of `DELETE` commands per minute. Operations reports high latency and `TombstoneOverwhelmingException` errors during reads. What architectural change should be made to resolve this issue? | pass→pass | 15,834 | 13,680 | -14% | 1 | 1 | 0% | 2,524 | 2,850 | +13% | 0 | 0 | — |
▸case-14 A backend service needs to fetch details for 25 distinct user records from Amazon DynamoDB, where each user record has a unique, non-sequential partition key. Developers are considering making 25 separate `GetItem` calls in a loop. What API method should be used instead to reduce network overhead? | fail→pass | 5,019 | 5,656 | +13% | 1 | 1 | 0% | 845 | 1,631 | +93% | 0 | 0 | — |
▸case-15 A banking application on Amazon DynamoDB needs to debit $100 from Account A's partition (`PK: ACC#A`) and credit $100 to Account B's partition (`PK: ACC#B`). The operations must either both succeed or both fail without risking partial updates. Which DynamoDB operation guarantees this cross-partition atomicity? | pass→pass | 4,497 | 27,135 | +503% | 1 | 1 | 0% | 833 | 1,378 | +65% | 0 | 0 | — |
▸case-16 In an Apache Cassandra cluster, temporary network glitches caused replicas to fall out of sync. To restore data consistency across all nodes without impacting live read latency or relying on read repairs during user traffic, what background operational utility should be executed? | pass→pass | 6,640 | 7,428 | +12% | 1 | 1 | 0% | 1,062 | 1,829 | +72% | 0 | 0 | — |
▸case-17 A DynamoDB table stores 300 KB PDF document metadata items, but queries frequently fetch only the `document_status` and `author_id` attributes. The team is seeing high Read Capacity Unit (RCU) consumption. How should the data storage architecture be reorganized to optimize cost? | pass→pass | 15,022 | 17,198 | +14% | 1 | 1 | 0% | 2,761 | 3,821 | +38% | 0 | 0 | — |
▸case-18 A developer wants to retrieve all orders placed by user `USR#789` from a DynamoDB table containing 10 million rows. They wrote a code snippet using `Scan` with a FilterExpression on `user_id = USR#789`. Why is this problematic and how should the request be performed? | pass→pass | 10,438 | 13,129 | +26% | 1 | 1 | 0% | 1,969 | 2,891 | +47% | 0 | 0 | — |
▸case-19 When creating a new CQL keyspace in Apache Cassandra across multiple data centers (e.g., `us-east` and `us-west`), developers defaulted to using `SimpleStrategy` with `replication_factor: 3`. What strategy class must be used instead for multi-datacenter deployments? | pass→pass | 3,430 | 4,279 | +25% | 1 | 1 | 0% | 547 | 1,314 | +140% | 0 | 0 | — |
▸case-20 An order management system on DynamoDB needs to retrieve order metadata (status, customer ID) along with all order line items (product, quantity) in a single request. How should sort key prefixes (`SK`) be structured within the order's partition key item collection (`PK: ORDER#1001`)? | pass→pass | 11,345 | 12,322 | +9% | 1 | 1 | 0% | 2,071 | 3,008 | +45% | 0 | 0 | — |
▸case-21 We are designing a PostgreSQL relational database for an enterprise ERP billing system. The team needs to eliminate transitive dependencies in customer address records where `zip_code` determines `city` and `state`. What database design process should be applied to structure these tables? | pass→pass | 13,567 | 10,361 | -24% | 1 | 1 | 0% | 2,550 | 2,445 | -4% | 0 | 0 | — |
▸case-22 A content management platform using MongoDB needs to store blog posts along with their nested user comments (up to 2 levels deep). The application always fetches a blog post together with all its comments in a single query. How should the schema be structured in MongoDB? | pass→fail | 16,244 | 17,679 | +9% | 1 | 1 | 0% | 2,594 | 3,303 | +27% | 0 | 0 | — |
▸case-23 An online gaming platform needs an in-memory caching tier to store real-time top-100 player scores. Low latency (< 2 ms) reads and writes are required, with automatic eviction when memory limits are reached. Which database system and native data structure best fits this requirement? | pass→pass | 9,563 | 9,668 | +1% | 1 | 1 | 0% | 1,795 | 2,359 | +31% | 0 | 0 | — |