PMLE Serving and Scaling Models Practice Question

A company uses Vertex AI Vector Search (Matching Engine) for a product recommendation system. The product embeddings are updated hourly. Which index update method should they use to ensure low latency for new items?

⚠ Common exam trap

A common mix-up: candidates assume batch rebuilds are the only reliable method for consistency, overlooking that streaming updates in Vertex AI Vector Search are designed specifically for low-latency incremental ingestion without sacrificing search quality.

Answer choices

Why each option matters

Answer the question above first, then reveal the full breakdown to understand why each option is right or wrong.

Correct answer & explanation

✓

Use streaming updates to add new embeddings incrementally

Vertex AI Vector Search supports streaming updates, allowing new embeddings to be added incrementally without rebuilding the entire index. This ensures low latency for new items by making them searchable almost immediately after update, which is critical for hourly refresh cycles where batch rebuilds would introduce significant delay.

Answer analysis

Option-by-option breakdown

For each option: why learners choose it and why it is or isn't the right answer here.

✗
Batch rebuild the index every hour
Why it's wrong here
Batch rebuilding the index every hour recreates the whole structure, causing latency spikes and leaving new items unsearchable between rebuilds. It is tempting because it is simple and matches the hourly cadence, and would suit embeddings that change infrequently where staleness is acceptable.
✓
Use streaming updates to add new embeddings incrementally
Why this is correct
Streaming updates let the index ingest new product embeddings incrementally without a full rebuild, so hourly-refreshed items become searchable almost immediately. This satisfies the low-latency requirement for new items, whereas batch rebuilds would delay their availability.
✗
Create a new index each hour and swap endpoints
Why it's wrong here
Rebuilding and swapping endpoints hourly forces a full reindex and re-upload, so new items only appear after each swap, not continuously. Streaming updates exist precisely to mutate a live index as embeddings change; batch rebuilds suit infrequent, wholesale embedding regeneration.
✗
Use brute-force index to simplify updates
Why it's wrong here
Brute-force search compares the query against every vector, so latency scales linearly with catalogue size and cannot meet low-latency serving. It suits small datasets or recall benchmarking, not hourly-refreshed production indexes where approximate nearest-neighbour structures are required.

Go deeper

PMLE Serving and Scaling Models Practice Question

Why each option matters

Option-by-option breakdown

Related to this question