SQL · Theory

Cluster

Theory

367/439

Cluster

Cluster is how bigger MySQL setups split or copy data. Replication: one primary takes writes, replicas copy via the binary log (binlog) for reads/backups. Sharding: split rows across servers by a key (app-level or extra tools). Partitioning: split one table into pieces on the same server (PARTITION BY RANGE …).

A fresher viva: ‘we replicate for backup/read scale; we shard only when one server cannot hold the data.’ Don’t design sharding on day one of a college project.

Trap: calling PARTITION the same as SHARD. Trap: replica as a second writable primary without a plan (see primary-primary only if they ask).

On the example next to this theory: Cluster: create two demo rows, then SELECT qty >= 2 ordered. Say which labels come back.

Exam tip

Replication vs partition vs shard in one line each.

Cluster — sample query

-- Cluster
CREATE TABLE demo (
  id INTEGER PRIMARY KEY,
  label TEXT NOT NULL,
  qty INTEGER DEFAULT 0
);
INSERT INTO demo (id, label, qty) VALUES
  (1, 'alpha', 2),
  (2, 'beta', 5);
SELECT label, qty FROM demo WHERE qty >= 2 ORDER BY qty DESC;

Cluster: create two demo rows, then SELECT qty >= 2 ordered. Say which labels come back.

Short notes

DefCluster — scale/copy strategy, not a SELECT keyword.
RuleReplication = copies. Partition = one table split. Shard = many servers.
TrapMixing those three words.

SQL · Theory

Cluster

← All stacks

Theory

367/439

Cluster

A fresher viva: ‘we replicate for backup/read scale; we shard only when one server cannot hold the data.’ Don’t design sharding on day one of a college project.

Trap: calling PARTITION the same as SHARD. Trap: replica as a second writable primary without a plan (see primary-primary only if they ask).

On the example next to this theory: Cluster: create two demo rows, then SELECT qty >= 2 ordered. Say which labels come back.

Exam tip

Replication vs partition vs shard in one line each.

Cluster — sample query

-- Cluster
CREATE TABLE demo (
  id INTEGER PRIMARY KEY,
  label TEXT NOT NULL,
  qty INTEGER DEFAULT 0
);
INSERT INTO demo (id, label, qty) VALUES
  (1, 'alpha', 2),
  (2, 'beta', 5);
SELECT label, qty FROM demo WHERE qty >= 2 ORDER BY qty DESC;

Cluster: create two demo rows, then SELECT qty >= 2 ordered. Say which labels come back.

Short notes

DefCluster — scale/copy strategy, not a SELECT keyword.
RuleReplication = copies. Partition = one table split. Shard = many servers.
TrapMixing those three words.

Cluster

Cluster — sample query

Short notes

Questions

Cluster

Cluster — sample query

Short notes

Questions