← Interview Preparation

Databases Reference

How data is stored — relational modeling, storage engines, transactions and MVCC, replication and sharding, NoSQL families, key-value and specialized stores, object/blob storage, data lakes and table formats, plus search and columnar OLAP. For SELECT/JOIN practice, use the SQL roadmap; for Raft, quorums, and CAP, use Distributed Systems. Primary textbooks: Designing Data-Intensive Applications (Kleppmann) and Database Internals (Petrov).

Time budget: ≈39h

How to use this reference

  • Work through topics top to bottom — storage internals and MVCC assume you understand relational schema shape; sharding assumes you understand replication.
  • DDIA and Database Internals are anchor books — each subtopic points to specific chapters, not cover-to-cover reading.
  • We deliberately do not repeat the SQL query-writing track, generic distributed-systems theory (Raft, sagas), or OS file-system deep dives — those are separate prerequisites or companions.
  • Scope is storage only — no Spark pipelines, Airflow orchestration, or stream processing. Data lakes and table formats are included because they define how files are organized and versioned on disk.

The Reference

  1. 1

    How relational data is shaped before it ever hits a query planner — normalization trade-offs, keys, and index design from a storage perspective.

    1. 1.1Schema Design & Normalization2/525m
    2. 1.2Keys, Constraints & Index Design3/530m
  2. 2

    What happens below the SQL layer — B-tree vs LSM engines, write-ahead logs, pages, and the buffer pool.

    1. 2.1B-Tree vs LSM Storage Engines4/530m
    2. 2.2WAL, Pages & Buffer Pool4/530m
    3. 2.3Bloom Filters3/530m
    4. 2.4External Sort & Large Queries3/525m
  3. 3

    ACID guarantees, isolation levels, and how Postgres-style MVCC actually stores row versions.

    1. 3.1ACID & Isolation Levels3/525m
    2. 3.2MVCC Visibility & Implementation4/530m
  4. 4

    How RDBMS products replicate data and fail over — the database lens on a problem whose theory lives in Distributed Systems.

    1. 4.1RDBMS Replication & Failover4/530m
  5. 5

    When one Postgres isn't enough — horizontal partitioning, shard keys, and how Vitess and Citus scale relational storage.

    1. 5.1Horizontal Sharding, Vitess & Citus4/530m
  6. 6

    Document and wide-column stores — when relational rows aren't the right storage primitive.

    1. 6.1Document Databases3/525m
    2. 6.2Wide-Column Stores4/530m
  7. 7

    Cache-aside through write-behind, stampede mitigation, and Redis, DynamoDB, graph, and time-series engines.

    1. 7.1Application Caching Patterns3/530m
    2. 7.2KV, Redis, DynamoDB, Graph & Time-Series4/535m
    3. 7.3HyperLogLog & Cardinality Estimation4/525m
  8. 8

    S3, HDFS, Parquet, and table formats — how analytical data is stored at petabyte scale without a traditional database server.

    1. 8.1Object & Blob Storage (S3, HDFS)3/525m
    2. 8.2Parquet & Lake Table Formats4/530m
  9. 9

    Inverted indexes for full-text search and columnar engines for analytics — two storage layouts optimized for very different read patterns.

    1. 9.1Elasticsearch & Columnar OLAP4/535m