Specialized Databases

By the end, you can

• Explain why embedding similar text near each other in a vector space enables semantic search.
• Describe why brute-force nearest-neighbor search is O(N · d) and how ANN indexes like HNSW address the scalability problem.
• Identify what RAG stands for and the role a vector database plays in it.
• Explain index-free adjacency and why it makes multi-hop graph traversals faster than stacked SQL JOINs.
• Distinguish Cypher (declarative pattern matching) from Gremlin (imperative traversal scripting).
• List the three primary use cases for graph databases covered in the lesson.
• Describe the write pattern that characterizes time-series data and why columnar + delta-of-delta + Gorilla compression achieve high compression ratios.
• Explain downsampling and retention policies and why chunk-drop expiry is cheaper than row-by-row deletion.
• Select the appropriate specialized database (vector, graph, or time-series) given a description of an access pattern, or choose relational when none fits.