Wide-column & LSM systems

How do writes scale across enormous sparse datasets?

Study partition-first data modeling, immutable files, compaction, tunable consistency, repair, and hotspot avoidance.

Systems compared

• Cassandra
• ScyllaDB
• HBase
• Bigtable

Open the engine

1. Memtables and SSTables
2. Compaction strategies
3. Partition and clustering keys
4. Quorums and repair
5. Tombstones and hotspots

Reconstruction lab

Design Cassandra tables from queries and diagnose a compaction or partition-key failure.