Integrate Azure Managed Redis in AI Solutions AI-200 Practice Questions (Page 6)

Questions 26–30

1. 26expert · hard

   A team is setting up vector search in Azure Managed Redis. They have a dataset of product embeddings that are updated daily. They need to support both exact and approximate search, and they want to minimize memory usage. Which approach should they take?

   • AUse only FLAT and accept higher latency for all queries.
   • BCreate two indexes: one FLAT for exact search and one HNSW for approximate search.
   • CUse only HNSW and accept approximate results for all queries.
   • DUse HNSW for approximate search and periodically rebuild a FLAT index for exact search only when needed.
   Reveal Solution Discussion Select an answer first
2. 27foundation · easy

   Which RediSearch command is used to perform a vector similarity search in Azure Managed Redis?

   • AFT.SEARCH
   • BFT.AGGREGATE
   • CSIM.SEARCH
   • DVECTOR.SEARCH
   Reveal Solution Discussion Select an answer first
4. 28expert · hard

   A media company uses Azure Managed Redis to store embeddings of video thumbnails for similarity search. The embeddings are updated frequently as new thumbnails are added. The team notices that search results are not reflecting newly added items. What is the most likely cause?

   • AThe TTL on the embeddings is too short, causing them to expire before search.
   • BThe new embeddings were stored in a different Redis database.
   • CThe vector index was not rebuilt after adding new embeddings.
   • DThe query is using a different distance metric than the index.
   Reveal Solution Discussion Select an answer first
5. 29expert · hard

   A healthcare application caches patient data in Azure Managed Redis. The data is sensitive and must not be served after a patient requests deletion (right to be forgotten). The application uses a cache-aside pattern. The team is debating between setting a short expiration (e.g., 5 minutes) versus deleting the cache key immediately upon a deletion request. Which approach should they choose?

   • AUse a write-through strategy to update the cache with a 'deleted' marker.
   • BDelete the cache key immediately when a deletion request is received, and also set a short expiration as a fallback.
   • CSet no expiration and rely on the cache eviction policy to remove old data.
   • DSet a 5-minute expiration on all patient data keys to ensure the data is eventually removed.
   Reveal Solution Discussion Select an answer first
6. 30application · medium

   A support-ticket system stores embeddings of past tickets in Azure Managed Redis. The team wants to find similar past tickets for a new incoming ticket by using cosine similarity. They have already created a vector index named 'ticket_embeddings' with the correct dimension and metric. What should they do to retrieve the top 5 most similar tickets for a given query embedding?

   • AUse the FT.AGGREGATE command with a GROUPBY on the embedding field and a REDUCE function to compute cosine similarity.
   • BUse the GET command with the query embedding as the key to retrieve similar tickets.
   • CUse the FT.SEARCH command with the query '(*)=>[KNN 5 @embedding $query_embedding AS score]' and provide the query embedding as a parameter.
   • DUse the SCAN command to iterate all keys and compute cosine similarity in the application code.
   Reveal Solution Discussion Select an answer first