CockroachDB The Definitive Guide, 2nd Edition (for Raymond Rhine) (Guy Harrison, Jesse Seldess, Ben Darnell etc.) (z-library.sk, 1lib.sk, z-lib.sk)

【One-Line Pitch】 A hands-on guide to building, tuning, and operating applications on CockroachDB, the distributed SQL database designed for elastic scale and resilience. Best for developers, architects, and DevOps/SRE teams who already know SQL and want to move from a local test cluster to production-grade, multi-region deployments. 【Book Arc】 - **Opening (~0%–12%)**: Establishes context — the evolution of databases toward distributed cloud SQL — then covers CockroachDB's architecture, its relationship to (and divergence from) Spanner, and where it fits versus analytic-only platforms. Solves the "what is this and when should I use it" question. - **Early (~12%–31%)**: Gets you running: local install, CockroachDB Cloud Basic, Docker/Kubernetes, certificates, sample workloads, and the SQL dialect (CTEs, recursive queries, column definitions, upserts). Then moves into schema design — keys, constraints, normalization vs. denormalization, and indexing. - **Middle (~31%–54%)**: Application development in practice: CRUD across language drivers, transaction patterns (optimistic retry vs. pessimistic `FOR UPDATE`), serializable isolation trade-offs, `AS OF SYSTEM TIME` reads, data import/migration from MySQL and PostgreSQL, and change data capture (changefeeds to Kafka, JSON/Avro). - **Late (~54%–end of excerpts)**: Query tuning and diagnostics — `EXPLAIN` output, full scans, join strategies (hash/merge/lookup) and how sensitive they are to data distribution. Excerpts do not cover the later deployment, backup, and disaster-recovery chapters in detail. - **Ending**: Part III is described as covering single- and multi-region deployment planning, backup, and disaster recovery, but the excerpts stop before this material, so treat those chapters as unverified here. 【Key Takeaways】 - **CockroachDB targets transactional workloads first** (Opening): it supports aggregations, windowing functions, and BI tools, but its differentiating features are aimed at transactions — analytic-only workloads may do better elsewhere. - **It is inspired by Spanner, not a clone** (Opening): notably, it does not assume special hardware like atomic clocks or GPS, which shapes its design choices. - **Ranges split on size and on load** (Opening): a range exceeding the ~512 MB threshold splits, and hot ranges can split earlier based on QPS thresholds, with the split point chosen to balance load. - **Getting started is deliberately low-friction** (Early): desktop install, free cloud tier, Docker, or Kubernetes, plus built-in sample schemas (bank, intro, kv, movr, startrek, tpcc) for practicing. - **Schema design favors artificial keys** (Early): they perform better, avoid the overhead of changing natural keys, and help distribute keys evenly across the cluster; foreign keys are often disabled in production for overhead reasons. - **Denormalization is a deliberate performance trade** (Early): reintroducing redundancy — e.g., storing foreign keys in an array instead of a join table — buys speed at the cost of normalization purity. - **Serializable isolation means conflicts are real** (Middle): CockroachDB's default is stricter than many SQL databases, so plan for retries (optimistic) or `FOR UPDATE` locking (pessimistic); retry logic is recommended either way. - **Read `EXPLAIN` for full scans** (Late): a `spans: FULL SCAN` often signals a missing index or a missing leading column in the `WHERE` clause, and join strategy choices are highly sensitive to data distribution. 【Reading Tips】 - Skim Part I's historical framing if you already know distributed SQL; slow down on the architecture and range-splitting sections, since they explain later behavior. - Treat the sample workloads (movr, tpcc, etc.) as your lab — run them locally before touching schema design or tuning chapters. - Deep-read the transaction and isolation material; it is the most common source of production surprises and the hardest to retrofit later. - Use the `EXPLAIN` and join-performance sections as a reference while tuning real queries rather than reading them linearly. - If you are an SRE, jump ahead to Part III's deployment/backup material once you have the schema and transaction basics. 【Coverage Limits】 This guide is based on stratified excerpts covering roughly the first half of the book; the deployment, backup, and disaster-recovery chapters in Part III are described but not substantively covered, so specifics there are not verified.