Struct SegmentWriterConfig 

pub struct SegmentWriterConfig {
    pub row_group_size: usize,
    pub max_row_group_size: usize,
    pub compress: bool,
    pub float_encoding: FloatEncoding,
    pub compression_codec: CompressionCodec,
    pub zstd_level: i32,
    pub column_codec_overrides: HashMap<String, CompressionCodec>,
    pub target_segment_size_bytes: usize,
    pub bloom_fpr: f64,
    pub zstd_dict_training: bool,
    pub field_encryption: FieldEncryptionConfig,
}

Configuration for the segment writer.

Fields

row_group_size: usize

Maximum rows per row group.

max_row_group_size: usize

Maximum rows per row group for memory-limiting purposes.

When set, the writer will split row groups to ensure that no single row group exceeds this number of rows. This bounds peak memory usage during encoding: only max_row_group_size rows worth of encoded column buffers are held in memory at any time.

Defaults to DEFAULT_ROW_GROUP_SIZE (65 536 rows). Reducing this trades write throughput (more row groups, more metadata) for lower peak memory consumption.

compress: bool

Whether to apply compression after encoding.

float_encoding: FloatEncoding

Float encoding strategy.

compression_codec: CompressionCodec

Compression codec (LZ4, Zstd, or None).

zstd_level: i32

Zstd compression level (1–22). Only used when compression_codec is Zstd.

column_codec_overrides: HashMap<String, CompressionCodec>

Per-column codec overrides. Keys are column names; values override the default compression_codec for that column.

target_segment_size_bytes: usize

Advisory maximum segment file size in bytes (default: 256 MiB).

When the estimated segment size exceeds this threshold, the flush logic in the database layer should split the data into multiple segments. This keeps individual segment files manageable for compaction, object-store upload, and cache eviction.

Note: This is an advisory limit — the writer itself produces a single segment file. Size-based splitting is enforced by the flush layer (chronix-memtable::flush::flush_frozen), which estimates the compressed row size and splits large measurement batches into multiple SegmentWriter instances.

bloom_fpr: f64

Bloom filter false-positive rate (default: 0.01 = 1%).

Lower values improve query pruning at the cost of larger bloom filter bitmaps. Recommended values: Low cardinality (<100 unique): 0.001 Default workloads: 0.01 High cardinality (>100K unique): 0.05

zstd_dict_training: bool

Enable Zstd dictionary training.

When true and compression_codec is Zstd, the writer collects encoded column blocks from the first row group as training samples, trains a dictionary via zstd::dict::from_samples, and compresses subsequent row groups with the trained dictionary. The dictionary is embedded in the segment metadata for the reader.

Most beneficial when row groups are small (< 64 KiB) and values contain repeating structure (e.g. string tags, enum fields).

field_encryption: FieldEncryptionConfig

Per-column field-level encryption configuration.

Only columns listed here are encrypted with AES-256-GCM. Encrypted columns have their statistics (min/max/sum/distinct) zeroed to prevent information leakage through zone maps.

Trait Implementations

impl Clone for SegmentWriterConfig

fn clone(&self) -> SegmentWriterConfig

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for SegmentWriterConfig

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl Default for SegmentWriterConfig

fn default() -> Self

Returns the “default value” for a type.