Skip to content
CQLite CQLite

Appendix E — Glossary

Concise definitions; cross-link to first use in the chapters.

  • SSTable: Immutable on-disk table used by Cassandra to store partitions and rows.

  • TOC (TOC.txt): Text file listing the set of components (Data/Index/Summary/etc.).

  • Data.db: Primary data file containing serialized partitions, rows, and cells.

  • Index.db: Partition index linking keys to Data.db offsets.

  • Summary.db: Sampled promoted index for faster index navigation.

  • Filter.db: Bloom filter for negative lookups.

  • Statistics.db: Table-level metadata including histograms and timestamps.

  • CompressionInfo.db: Chunk sizes, offsets, and checksums for compressed Data.db.

  • Digest (Digest.crc32): File CRC for integrity verification.

  • VInt: Variable-length integer encoding used for lengths and counters.

  • ZigZag: Encoding mapping signed integers to small unsigned values.

  • UDT: User-defined type in CQL.

  • SAI: Storage-Attached Index; secondary indexing subsystem (incl. vector).

  • BTI: B-Tree/Trie indexed SSTable format family in Cassandra 5.x. See 17-bti-formats.md.

  • TrieMemtable: An opt-in memtable in Cassandra 5.0, using a byte-ordered prefix trie (InMemoryTrie) for efficient memory usage and reduced GC pressure. Stores partitions using ByteComparable keys with prefix sharing; sharded across CPU cores for write concurrency. The default memtable in 5.0 is SkipListMemtable (MemtableParams.java:99); opt in via memtable: { class: TrieMemtable } in the table schema or cassandra.yaml. See 04-from-cql-to-disk.md and TrieMemtable.java.

  • big format: Legacy SSTable format family (pre-BTI) with classic BigTableReader/Writer. See 02-anatomy-of-an-sstable.md and 17-bti-formats.md for contrasts.

  • LCS/STCS/TWCS: Compaction strategies. See 15-compaction-strategies.md.

References:

  • Cassandra 5.0 source packages: see references/source-map.md.