TidesDB C API Reference

If you want to download the source of this document, you can find it here.

---

Want to watch a dedicated how-to video?

Overview

TidesDB is designed to provide a simple and intuitive C API for all your embedded storage needs. This document is complete reference for the C API covering database operations, transactions, column families, iterators and more.

Include

Note

You can use other components of TidesDB such as skip list, bloom filter etc. under tidesdb/ - this also prevents collisions.

Choosing Between tidesdb.h and db.h

TidesDB provides two header files with different purposes:

tidesdb.h

Full C implementation header. Mainly use this for native C/C++ applications.

#include <tidesdb/tidesdb.h>

db.h

db.h is mainly an FFI/Language binding interface with minimal dependencies and simpler ABI.

#include <tidesdb/db.h>

When to Use Each

• C/C++ applications ➞ Use tidesdb.h for full access to the API
• Language bindings (jextract, rust-bindgen, ctypes, cgo, etc.) ➞ Use db.h for a stable FFI interface
• Library developers ➞ Use db.h to avoid exposing internal implementation details

Error Codes

TidesDB provides detailed error codes for production use.

Code	Value	Description
TDB_SUCCESS	0	Operation completed successfully
TDB_ERR_MEMORY	-1	Memory allocation failed
TDB_ERR_INVALID_ARGS	-2	Invalid arguments passed to function (NULL pointers, invalid sizes, etc.)
TDB_ERR_NOT_FOUND	-3	Key not found in column family
TDB_ERR_IO	-4	I/O operation failed (file read/write error)
TDB_ERR_CORRUPTION	-5	Data corruption detected (checksum failure, invalid format version, truncated data)
TDB_ERR_EXISTS	-6	Resource already exists (e.g., column family name collision)
TDB_ERR_CONFLICT	-7	Transaction conflict detected (write-write or read-write conflict in SERIALIZABLE/SNAPSHOT isolation)
TDB_ERR_TOO_LARGE	-8	Key or value size exceeds maximum allowed size
TDB_ERR_MEMORY_LIMIT	-9	Operation would exceed memory limits (safety check to prevent OOM)
TDB_ERR_INVALID_DB	-10	Database handle is invalid (e.g., after close)
TDB_ERR_UNKNOWN	-11	Unknown or unspecified error
TDB_ERR_LOCKED	-12	Database is locked by another process

Error categories

• TDB_ERR_CORRUPTION indicates data integrity issues requiring immediate attention
• TDB_ERR_CONFLICT indicates transaction conflicts (retry may succeed)
• TDB_ERR_MEMORY, TDB_ERR_MEMORY_LIMIT, TDB_ERR_TOO_LARGE indicate resource constraints
• TDB_ERR_NOT_FOUND, TDB_ERR_EXISTS are normal operational conditions, not failures

Example Error Handling

int result = tidesdb_txn_put(txn, cf, key, key_size, value, value_size, -1);
if (result != TDB_SUCCESS)
{
    switch (result)
    {
        case TDB_ERR_MEMORY:
            fprintf(stderr, "out of memory\n");
            break;
        case TDB_ERR_INVALID_ARGS:
            fprintf(stderr, "invalid arguments\n");
            break;
        case TDB_ERR_CONFLICT:
            fprintf(stderr, "transaction conflict detected\n");
            break;
        default:
            fprintf(stderr, "operation failed with error code: %d\n", result);
            break;
    }
    return -1;
}

Initialization

TidesDB supports optional custom memory allocators for integration with custom memory managers (e.g., a Redis module allocator, jemalloc, etc). You don’t need to do this, this is completely optional.

tidesdb_init

Initializes TidesDB with optional custom memory allocation functions. When in use, this must be called exactly once before any other TidesDB function.

int tidesdb_init(tidesdb_malloc_fn malloc_fn, tidesdb_calloc_fn calloc_fn,
                 tidesdb_realloc_fn realloc_fn, tidesdb_free_fn free_fn);

Parameters

Name	Type	Description
malloc_fn	tidesdb_malloc_fn	Custom malloc function (or NULL for system malloc)
calloc_fn	tidesdb_calloc_fn	Custom calloc function (or NULL for system calloc)
realloc_fn	tidesdb_realloc_fn	Custom realloc function (or NULL for system realloc)
free_fn	tidesdb_free_fn	Custom free function (or NULL for system free)

Returns

• 0 on success
• -1 if already initialized

Example (system allocator)

tidesdb_init(NULL, NULL, NULL, NULL);

Example (custom allocator)

tidesdb_init(RedisModule_Alloc, RedisModule_Calloc,
             RedisModule_Realloc, RedisModule_Free);

Auto-initialization

If tidesdb_init() is not called, TidesDB will auto-initialize with the system allocator on the first call to tidesdb_open().

tidesdb_finalize

Finalizes TidesDB and resets the allocator. Should be called after all TidesDB operations are complete. After calling this, tidesdb_init() can be called again.

void tidesdb_finalize(void);

Example

tidesdb_init(NULL, NULL, NULL, NULL);

/* ... use TidesDB ... */

tidesdb_close(db);
tidesdb_finalize();

Storage Engine Operations

Opening TidesDB

tidesdb_config_t config = {
    .db_path = "./mydb",
    .num_flush_threads = 2,                /* Flush thread pool size (default: 2) */
    .num_compaction_threads = 2,           /* Compaction thread pool size (default: 2) */
    .log_level = TDB_LOG_INFO,             /* Log level: TDB_LOG_DEBUG, TDB_LOG_INFO, TDB_LOG_WARN, TDB_LOG_ERROR, TDB_LOG_FATAL, TDB_LOG_NONE */
    .block_cache_size = 64 * 1024 * 1024,  /* 64MB global block cache (default: 64MB) */
    .max_open_sstables = 256,              /* Max cached SSTable structures (default: 256) */
    .log_to_file = 0,                      /* Write logs to file instead of stderr (default: 0) */
    .log_truncation_at = 24 * (1024*1024), /* Log file truncation size (default: 24MB), 0 = no truncation */
};

tidesdb_t *db = NULL;
if (tidesdb_open(&config, &db) != 0)
{
    return -1;
}

if (tidesdb_close(db) != 0)
{
    return -1;
}

Using default configuration

Use tidesdb_default_config() to get a configuration with sensible defaults, then override specific fields as needed:

tidesdb_config_t config = tidesdb_default_config();
config.db_path = "./mydb";
config.log_level = TDB_LOG_WARN;  /* Override: only warnings and errors */

tidesdb_t *db = NULL;
if (tidesdb_open(&config, &db) != 0)
{
    return -1;
}

Multiple DBs Allowed

Multiple TidesDB instances can be opened in the same process, each with its own configuration and data directory.

Logging

TidesDB provides structured logging with multiple severity levels.

Log Levels

• TDB_LOG_DEBUG · Detailed diagnostic information
• TDB_LOG_INFO · General informational messages (default)
• TDB_LOG_WARN · Warning messages for potential issues
• TDB_LOG_ERROR · Error messages for failures
• TDB_LOG_FATAL · Critical errors that may cause shutdown
• TDB_LOG_NONE · Disable all logging

Configure at startup

tidesdb_config_t config = {
    .db_path = "./mydb",
    .log_level = TDB_LOG_DEBUG  /* Enable debug logging */
};

tidesdb_t *db = NULL;
tidesdb_open(&config, &db);

Production configuration

tidesdb_config_t config = {
    .db_path = "./mydb",
    .log_level = TDB_LOG_WARN  /* Only warnings and errors */
};

Output format Logs are written to stderr by default with timestamps:

[HH:MM:SS.mmm] [LEVEL] filename:line: message

Example output

[22:58:00.454] [INFO] tidesdb.c:9322: Opening TidesDB with path=./mydb
[22:58:00.456] [INFO] tidesdb.c:9478: Block clock cache created with max_bytes=64.00 MB

Log to file

Enable log_to_file to write logs to a LOG file in the database directory instead of stderr:

tidesdb_config_t config = {
    .db_path = "./mydb",
    .log_level = TDB_LOG_DEBUG,
    .log_to_file = 1  /* Write to ./mydb/LOG instead of stderr */
};

tidesdb_t *db = NULL;
tidesdb_open(&config, &db);
/* Logs are now written to ./mydb/LOG */

The log file is opened in append mode and uses line buffering for real-time logging. If the log file cannot be opened, logging falls back to default.

Redirect stderr to file (alternative)

./your_program 2> tidesdb.log  # Redirect std output to file

Backup

tidesdb_backup creates an on-disk snapshot of an open database without blocking normal reads/writes.

int tidesdb_backup(tidesdb_t *db, char *dir);

Usage

tidesdb_t *db = NULL;
tidesdb_open(&config, &db);

if (tidesdb_backup(db, "./mydb_backup") != 0)
{
    fprintf(stderr, "Backup failed\n");
}

Behavior

• Requires dir to be a non-existent directory or an empty directory; returns TDB_ERR_EXISTS if not empty.
• Does not copy the LOCK file, so the backup can be opened normally.
• Two-phase copy approach:
  • Copies immutable files first (SSTables listed in the manifest plus metadata/config files) and skips WALs.
  • Forces memtable flushes, waits for flush/compaction queues to drain, then copies any remaining files (including WALs and updated manifests). Existing SSTables already copied are not recopied.
• Database stays open and usable during backup; no exclusive lock is taken on the source directory.

Notes

• The backup represents the database state after the final flush/compaction drain.
• If you need a quiesced backup window, you can pause writes at the application level before calling this API.

Checkpoint

tidesdb_checkpoint creates a lightweight, near-instant snapshot of an open database using hard links instead of copying SSTable data.

int tidesdb_checkpoint(tidesdb_t *db, const char *checkpoint_dir);

Usage

tidesdb_t *db = NULL;
tidesdb_open(&config, &db);

if (tidesdb_checkpoint(db, "./mydb_checkpoint") != 0)
{
    fprintf(stderr, "Checkpoint failed\n");
}

Behavior

• Requires checkpoint_dir to be a non-existent directory or an empty directory; returns TDB_ERR_EXISTS if not empty.
• For each column family:
  • Flushes the active memtable so all data is in SSTables.
  • Halts compactions to ensure a consistent view of live SSTable files.
  • Hard links all SSTable files (.klog and .vlog) into the checkpoint directory, preserving the level subdirectory structure.
  • Copies small metadata files (manifest, config) into the checkpoint directory.
  • Resumes compactions.
• Falls back to file copy if hard linking fails (e.g., cross-filesystem).
• Database stays open and usable during checkpoint; no exclusive lock is taken on the source directory.

Checkpoint vs Backup

	tidesdb_backup	tidesdb_checkpoint
Speed	Copies every SSTable byte-by-byte	Near-instant (hard links, O(1) per file)
Disk usage	Full independent copy	No extra disk until compaction removes old SSTables
Portability	Can be moved to another filesystem or machine	Same filesystem only (hard link requirement)
Use case	Archival, disaster recovery, remote shipping	Fast local snapshots, point-in-time reads, streaming backups

Notes

• The checkpoint represents the database state at the point all memtables are flushed and compactions are halted.
• Hard-linked files share storage with the live database. Deleting the original database does not affect the checkpoint (hard link semantics).
• The checkpoint can be opened as a normal TidesDB database with tidesdb_open.

Column Family Operations

Creating a Column Family

Column families are isolated key-value stores. Use the config struct for customization or use defaults.

/* Create with default configuration */
tidesdb_column_family_config_t cf_config = tidesdb_default_column_family_config();

if (tidesdb_create_column_family(db, "my_cf", &cf_config) != 0)
{
    return -1;
}

Custom configuration example

tidesdb_column_family_config_t cf_config = {
    .write_buffer_size = 128 * 1024 * 1024,     /* 128MB memtable flush threshold */
    .level_size_ratio = 10,                     /* Level size multiplier (default: 10) */
    .min_levels = 5,                            /* Minimum LSM levels (default: 5) */
    .dividing_level_offset = 2,                 /* Compaction dividing level offset (default: 2) */
    .skip_list_max_level = 12,                  /* Skip list max level */
    .skip_list_probability = 0.25f,             /* Skip list probability */
    .compression_algorithm = TDB_COMPRESS_LZ4,  /* TDB_COMPRESS_LZ4, TDB_COMPRESS_LZ4_FAST, TDB_COMPRESS_ZSTD, TDB_COMPRESS_SNAPPY, or TDB_COMPRESS_NONE */
    .enable_bloom_filter = 1,                   /* Enable bloom filters */
    .bloom_fpr = 0.01,                          /* 1% false positive rate */
    .enable_block_indexes = 1,                  /* Enable compact block indexes */
    .index_sample_ratio = 1,                    /* Sample every block for index (default: 1) */
    .block_index_prefix_len = 16,               /* Block index prefix length (default: 16) */
    .sync_mode = TDB_SYNC_FULL,                 /* TDB_SYNC_NONE, TDB_SYNC_INTERVAL, or TDB_SYNC_FULL */
    .sync_interval_us = 1000000,                /* Sync interval in microseconds (1 second, only for TDB_SYNC_INTERVAL) */
    .comparator_name = {0},                     /* Empty = use default "memcmp" */
    .klog_value_threshold = 512,                /* Values > 512 bytes go to vlog (default: 512) */
    .min_disk_space = 100 * 1024 * 1024,        /* Minimum disk space required (default: 100MB) */
    .default_isolation_level = TDB_ISOLATION_READ_COMMITTED,  /* Default transaction isolation */
    .l1_file_count_trigger = 4,                 /* L1 file count trigger for compaction (default: 4) */
    .l0_queue_stall_threshold = 20,             /* L0 queue stall threshold (default: 20) */
    .use_btree = 0                              /* Use B+tree format for klog (default: 0 = block-based) */
};

if (tidesdb_create_column_family(db, "my_cf", &cf_config) != 0)
{
    return -1;
}

Using custom comparator

/* Register comparator after opening database but before creating CF */
tidesdb_register_comparator(db, "reverse", my_reverse_compare, NULL, NULL);

tidesdb_column_family_config_t cf_config = tidesdb_default_column_family_config();
strncpy(cf_config.comparator_name, "reverse", TDB_MAX_COMPARATOR_NAME - 1);
cf_config.comparator_name[TDB_MAX_COMPARATOR_NAME - 1] = '\0';

if (tidesdb_create_column_family(db, "sorted_cf", &cf_config) != 0)
{
    return -1;
}

Dropping a Column Family

By name (looks up column family internally):

if (tidesdb_drop_column_family(db, "my_cf") != 0)
{
    return -1;
}

By pointer (skips name lookup when you already have the pointer):

tidesdb_column_family_t *cf = tidesdb_get_column_family(db, "my_cf");
if (!cf) return -1;

if (tidesdb_delete_column_family(db, cf) != 0)
{
    return -1;
}

Which to use

• tidesdb_drop_column_family(db, name) · convenient when you only have the name
• tidesdb_delete_column_family(db, cf) · faster when you already hold a tidesdb_column_family_t*, avoids a redundant linear scan

Renaming a Column Family

Atomically rename a column family and its underlying directory. The operation waits for any in-progress flush or compaction to complete before renaming.

if (tidesdb_rename_column_family(db, "old_name", "new_name") != 0)
{
    return -1;
}

Behavior

• Waits for any in-progress flush or compaction to complete
• Atomically renames the column family directory on disk
• Updates all internal paths (SSTables, manifest, config)
• Thread-safe with proper locking

Return values

• TDB_SUCCESS · Rename completed successfully
• TDB_ERR_NOT_FOUND · Column family with old_name doesn’t exist
• TDB_ERR_EXISTS · Column family with new_name already exists
• TDB_ERR_IO · Failed to rename directory on disk

Cloning a Column Family

Create a complete copy of an existing column family with a new name. The clone contains all the data from the source at the time of cloning.

if (tidesdb_clone_column_family(db, "source_cf", "cloned_cf") != 0)
{
    return -1;
}

/* Both column families now exist independently */
tidesdb_column_family_t *original = tidesdb_get_column_family(db, "source_cf");
tidesdb_column_family_t *clone = tidesdb_get_column_family(db, "cloned_cf");

Behavior

• Flushes the source column family’s memtable to ensure all data is on disk
• Waits for any in-progress flush or compaction to complete
• Copies all SSTable files (.klog and .vlog) to the new directory
• Copies manifest and configuration files
• Creates a new column family structure and loads the copied SSTables
• The clone is completely independent - modifications to one do not affect the other

Use cases

• Testing · Create a copy of production data for testing without affecting the original
• Branching · Create a snapshot of data before making experimental changes
• Migration · Clone data before schema or configuration changes
• Backup verification · Clone and verify data integrity without modifying the source

Return values

• TDB_SUCCESS · Clone completed successfully
• TDB_ERR_NOT_FOUND · Source column family doesn’t exist
• TDB_ERR_EXISTS · Destination column family already exists
• TDB_ERR_INVALID_ARGS · Invalid arguments (NULL pointers or same source/destination name)
• TDB_ERR_IO · Failed to copy files or create directory

Clone vs Backup

tidesdb_clone_column_family creates a new column family within the same database instance. For creating an external backup of the entire database, use tidesdb_backup instead.

Getting a Column Family

Retrieve a column family pointer to use in operations.

tidesdb_column_family_t *cf = tidesdb_get_column_family(db, "my_cf");
if (cf == NULL)
{
    /* Column family not found */
    return -1;
}

Listing Column Families

Get all column family names on the TidesDB instance.

char **names = NULL;
int count = 0;

if (tidesdb_list_column_families(db, &names, &count) == 0)
{
    printf("Found %d column families:\n", count);
    for (int i = 0; i < count; i++)
    {
        printf("  - %s\n", names[i]);
        free(names[i]);
    }
    free(names);
}

Column Family Statistics

Get detailed statistics about a column family.

tidesdb_column_family_t *cf = tidesdb_get_column_family(db, "my_cf");
if (!cf) return -1;

tidesdb_stats_t *stats = NULL;
if (tidesdb_get_stats(cf, &stats) == 0)
{
    printf("Memtable Size: %zu bytes\n", stats->memtable_size);
    printf("Number of Levels: %d\n", stats->num_levels);
    printf("Total Keys: %" PRIu64 "\n", stats->total_keys);
    printf("Total Data Size: %" PRIu64 " bytes\n", stats->total_data_size);
    printf("Avg Key Size: %.1f bytes\n", stats->avg_key_size);
    printf("Avg Value Size: %.1f bytes\n", stats->avg_value_size);
    printf("Read Amplification: %.2f\n", stats->read_amp);
    printf("Cache Hit Rate: %.1f%%\n", stats->hit_rate * 100.0);

    /* Column family name is available via config */
    printf("Column Family: %s\n", stats->config->name);

    for (int i = 0; i < stats->num_levels; i++)
    {
        printf("Level %d: %d SSTables, %zu bytes, %" PRIu64 " keys\n",
               i + 1, stats->level_num_sstables[i], stats->level_sizes[i],
               stats->level_key_counts[i]);
    }

    /* B+tree stats (only populated if use_btree=1) */
    if (stats->use_btree)
    {
        printf("B+tree Total Nodes: %" PRIu64 "\n", stats->btree_total_nodes);
        printf("B+tree Max Height: %u\n", stats->btree_max_height);
        printf("B+tree Avg Height: %.2f\n", stats->btree_avg_height);
    }

    /* Access configuration */
    printf("Write Buffer Size: %zu\n", stats->config->write_buffer_size);
    printf("Compression: %d\n", stats->config->compression_algorithm);
    printf("Bloom Filter: %s\n", stats->config->enable_bloom_filter ? "enabled" : "disabled");

    tidesdb_free_stats(stats);
}

Statistics include

Field	Type	Description
num_levels	int	Number of LSM levels
memtable_size	size_t	Current memtable size in bytes
level_sizes	size_t*	Array of per-level total sizes
level_num_sstables	int*	Array of per-level SSTable counts
level_key_counts	uint64_t*	Array of per-level key counts
config	tidesdb_column_family_config_t*	Full column family configuration. This includes column family name if you need it!
total_keys	uint64_t	Total keys across memtable and all SSTables
total_data_size	uint64_t	Total data size (klog + vlog) in bytes
avg_key_size	double	Estimated average key size in bytes
avg_value_size	double	Estimated average value size in bytes
read_amp	double	Read amplification factor (point lookup cost)
hit_rate	double	Block cache hit rate (0.0 to 1.0)
use_btree	int	Whether column family uses B+tree format
btree_total_nodes	uint64_t	Total B+tree nodes across all SSTables
btree_max_height	uint32_t	Maximum tree height across all SSTables
btree_avg_height	double	Average tree height across all SSTables

B+tree Statistics

The B+tree stats (btree_total_nodes, btree_max_height, btree_avg_height) are only populated when use_btree=1 in the column family configuration. These provide insight into the index structure overhead and lookup depth.

Block Cache Statistics

Get statistics for the global block cache (shared across all column families).

tidesdb_cache_stats_t cache_stats;
if (tidesdb_get_cache_stats(db, &cache_stats) == 0)
{
    if (cache_stats.enabled)
    {
        printf("Cache enabled: yes\n");
        printf("Total entries: %zu\n", cache_stats.total_entries);
        printf("Total bytes: %.2f MB\n", cache_stats.total_bytes / (1024.0 * 1024.0));
        printf("Hits: %lu\n", cache_stats.hits);
        printf("Misses: %lu\n", cache_stats.misses);
        printf("Hit rate: %.1f%%\n", cache_stats.hit_rate * 100.0);
        printf("Partitions: %zu\n", cache_stats.num_partitions);
    }
    else
    {
        printf("Cache enabled: no (block_cache_size = 0)\n");
    }
}

Cache statistics include

• enabled · Whether block cache is active (0 if block_cache_size was set to 0)
• total_entries · Number of cached blocks
• total_bytes · Total memory used by cached blocks
• hits · Number of cache hits (blocks served from memory)
• misses · Number of cache misses (blocks read from disk)
• hit_rate · Hit rate as a decimal (0.0 to 1.0)
• num_partitions · Number of cache partitions (scales with CPU cores)

Block Cache

The block cache is a database-level resource shared across all column families. It caches deserialized klog blocks to avoid repeated disk I/O and deserialization. Configure cache size via config.block_cache_size when opening the database. Set to 0 to disable caching.

Range Cost Estimation

tidesdb_range_cost estimates the computational cost of iterating between two keys in a column family. The returned value is an opaque double — meaningful only for comparison with other values from the same function. It uses only in-memory metadata and performs no disk I/O.

int tidesdb_range_cost(tidesdb_column_family_t *cf,
                       const uint8_t *key_a, size_t key_a_size,
                       const uint8_t *key_b, size_t key_b_size,
                       double *cost);

Parameters

Name	Type	Description
cf	tidesdb_column_family_t*	Column family to estimate cost for
key_a	const uint8_t*	First key (bound of range)
key_a_size	size_t	Size of first key
key_b	const uint8_t*	Second key (bound of range)
key_b_size	size_t	Size of second key
cost	double*	Output: estimated traversal cost (higher = more expensive)

Returns

• TDB_SUCCESS on success
• TDB_ERR_INVALID_ARGS on bad input (NULL pointers, zero-length keys)

Example

tidesdb_column_family_t *cf = tidesdb_get_column_family(db, "my_cf");
if (!cf) return -1;

double cost_a = 0.0, cost_b = 0.0;

tidesdb_range_cost(cf, (uint8_t *)"user:0000", 9,
                       (uint8_t *)"user:0999", 9, &cost_a);

tidesdb_range_cost(cf, (uint8_t *)"user:1000", 9,
                       (uint8_t *)"user:1099", 9, &cost_b);

if (cost_a < cost_b)
{
    printf("Range A is cheaper to iterate\n");
}

How it works

The function walks all SSTable levels and uses in-memory metadata to estimate how many blocks and entries fall within the given key range:

• With block indexes enabled · Uses O(log B) binary search per overlapping SSTable to find the block slots containing each key bound. The block span between slots, scaled by index_sample_ratio, gives the estimated block count.
• Without block indexes · Falls back to byte-level key interpolation. The leading 8 bytes of each key are converted to a numeric position within the SSTable’s min/max key range to estimate the fraction of blocks covered.
• B+tree SSTables (use_btree=1) · Uses the same key interpolation against tree node counts, plus tree height as a seek cost. Only applies to column families configured with B+tree klog format.
• Compression · Compressed SSTables receive a 1.5× weight multiplier to account for decompression overhead.
• Merge overhead · Each overlapping SSTable adds a small fixed cost for merge-heap operations.
• Memtable · The active memtable’s entry count contributes a small in-memory cost.

Key order does not matter — the function normalizes the range so key_a > key_b produces the same result as key_b > key_a.

Use cases

• Query planning · Compare candidate key ranges to find the cheapest one to scan
• Load balancing · Distribute range scan work across threads by estimating per-range cost
• Adaptive prefetching · Decide how aggressively to prefetch based on range size
• Monitoring · Track how data distribution changes across key ranges over time

Cost Values

The returned cost is not an absolute measure (it does not represent milliseconds, bytes, or entry counts). It is a relative scalar — only meaningful when compared with other tidesdb_range_cost results. A cost of 0.0 means no overlapping SSTables or memtable entries were found for the range.

Compression Algorithms

TidesDB supports multiple compression algorithms to reduce storage footprint and I/O bandwidth. Compression is applied to both klog (key-log) and vlog (value-log) blocks before writing to disk.

Available Algorithms

• TDB_COMPRESS_NONE · No compression (value: 0)

  • Raw data written directly to disk
  • Use case · Pre-compressed data, maximum write throughput, CPU-constrained environments

• TDB_COMPRESS_LZ4 · LZ4 standard compression (value: 2, default)

  • Fast compression and decompression with good compression ratios
  • Use case · General purpose, balanced performance and compression
  • Performance · ~500 MB/s compression, ~2000 MB/s decompression (typical)

• TDB_COMPRESS_LZ4_FAST · LZ4 fast mode (value: 4)

  • Faster compression than standard LZ4 with slightly lower compression ratio
  • Uses acceleration factor of 2
  • Use case · Write-heavy workloads prioritizing speed over compression ratio
  • Performance · Higher compression throughput than standard LZ4

• TDB_COMPRESS_ZSTD · Zstandard compression (value: 3)

  • Best compression ratio with moderate speed (compression level 1)
  • Use case · Storage-constrained environments, archival data, read-heavy workloads
  • Performance · ~400 MB/s compression, ~1000 MB/s decompression (typical)

• TDB_COMPRESS_SNAPPY · Snappy compression (value: 1)

  • Fast compression with moderate compression ratios
  • Availability · Not available on SunOS/Illumos/OmniOS platforms
  • Use case · Legacy compatibility, platforms where Snappy is preferred

Configuration Example

tidesdb_column_family_config_t cf_config = tidesdb_default_column_family_config();

/* Use LZ4 compression (default) */
cf_config.compression_algorithm = TDB_COMPRESS_LZ4;

/* Use Zstandard for better compression ratio */
cf_config.compression_algorithm = TDB_COMPRESS_ZSTD;

/* Use LZ4 fast mode for maximum write throughput */
cf_config.compression_algorithm = TDB_COMPRESS_LZ4_FAST;

/* Disable compression */
cf_config.compression_algorithm = TDB_COMPRESS_NONE;

tidesdb_create_column_family(db, "my_cf", &cf_config);

Important

Compression algorithm cannot be changed after column family creation without corrupting existing SSTables. Compression is applied at the block level (both klog and vlog blocks).

• Decompression happens automatically during reads
• Block cache stores decompressed blocks to avoid repeated decompression overhead
• Different column families can use different compression algorithms

B+tree KLog Format (Optional)

Column families can optionally use a B+tree structure for the key log instead of the default block-based format. The B+tree format offers faster point lookups through O(log N) tree traversal rather than linear block scanning.

Enabling B+tree format

tidesdb_column_family_config_t cf_config = tidesdb_default_column_family_config();
cf_config.use_btree = 1;  /* Enable B+tree format */

tidesdb_create_column_family(db, "btree_cf", &cf_config);

Characteristics

• Point lookups · O(log N) tree traversal with binary search at each node, compared to potentially scanning multiple 64KB blocks in block-based format
• Range scans · Doubly-linked leaf nodes enable efficient bidirectional iteration
• Immutable · Tree is bulk-loaded from sorted memtable data during flush and never modified afterward
• Compression · Nodes compress independently using the same algorithms (LZ4, LZ4-FAST, Zstd)
• Large values · Values exceeding klog_value_threshold are stored in vlog, same as block-based format
• Bloom filter · Works identically — checked before tree traversal to skip lookups for absent keys

When to use B+tree format

• Read-heavy workloads with frequent point lookups
• Workloads where read latency is more important than write throughput
• Large SSTables where block scanning becomes expensive

Tradeoffs

• Slightly higher write amplification during flush (building tree structure)
• Larger metadata overhead per node compared to block-based format
• Block-based format may be faster for sequential scans of entire SSTables

Important

use_btree cannot be changed after column family creation. Different column families can use different formats (some B+tree, some block-based). Both formats support the same compression algorithms and bloom filters.

Choosing a Compression Algorithm

Workload	Recommended Algorithm	Rationale
General purpose	TDB_COMPRESS_LZ4	Best balance of speed and compression
Write-heavy	TDB_COMPRESS_LZ4_FAST	Minimize CPU overhead on writes
Storage-constrained	TDB_COMPRESS_ZSTD	Maximum compression ratio
Read-heavy	TDB_COMPRESS_ZSTD	Reduce I/O bandwidth, decompression is fast
Pre-compressed data	TDB_COMPRESS_NONE	Avoid double compression overhead
CPU-constrained	TDB_COMPRESS_NONE or TDB_COMPRESS_LZ4_FAST	Minimize CPU usage

Updating Column Family Configuration

Update runtime-safe configuration settings. Configuration changes are applied to new operations only.

tidesdb_column_family_t *cf = tidesdb_get_column_family(db, "my_cf");
if (!cf) return -1;

tidesdb_column_family_config_t new_config = tidesdb_default_column_family_config();
new_config.write_buffer_size = 256 * 1024 * 1024;
new_config.skip_list_max_level = 16;
new_config.skip_list_probability = 0.25f;
new_config.bloom_fpr = 0.001;       /* 0.1% false positive rate */
new_config.index_sample_ratio = 8;  /* sample 1 in 8 keys */

int persist_to_disk = 1;            /* save to config.ini */
if (tidesdb_cf_update_runtime_config(cf, &new_config, persist_to_disk) == 0)
{
    printf("Configuration updated successfully\n");
}

Updatable settings (safe to change at runtime):

• write_buffer_size · Memtable flush threshold
• skip_list_max_level · Skip list level for new memtables
• skip_list_probability · Skip list probability for new memtables
• bloom_fpr · False positive rate for new SSTables
• index_sample_ratio · Index sampling ratio for new SSTables
• sync_mode · Durability mode (TDB_SYNC_NONE, TDB_SYNC_INTERVAL, or TDB_SYNC_FULL)
• sync_interval_us · Sync interval in microseconds (only used when sync_mode is TDB_SYNC_INTERVAL)

Non-updatable settings (would corrupt existing data):

• compression_algorithm · Cannot change on existing SSTables
• enable_block_indexes · Cannot change index structure
• enable_bloom_filter · Cannot change bloom filter presence
• comparator_name · Cannot change sort order
• level_size_ratio · Cannot change LSM level sizing
• klog_value_threshold · Cannot change klog/vlog separation
• min_levels · Cannot change minimum LSM levels
• dividing_level_offset · Cannot change compaction strategy
• block_index_prefix_len · Cannot change block index structure
• l1_file_count_trigger · Cannot change compaction trigger
• l0_queue_stall_threshold · Cannot change backpressure threshold
• use_btree · Cannot change klog format after creation

Backpressure Defaults

The default l0_queue_stall_threshold is 20. The default l1_file_count_trigger is 4.

Configuration persistence

If persist_to_disk = 1, changes are saved to config.ini in the column family directory. On restart, the configuration is loaded from this file.

/* Save configuration to custom INI file */
tidesdb_cf_config_save_to_ini("custom_config.ini", "my_cf", &new_config);

/* Load configuration from INI file */
tidesdb_column_family_config_t loaded_config;
if (tidesdb_cf_config_load_from_ini("custom_config.ini", "my_cf", &loaded_config) == 0)
{
    printf("Configuration loaded successfully\n");
}

Important Notes

Changes apply immediately to new operations. Existing SSTables and memtables retain their original settings. The update operation is thread-safe.

Transactions

All operations in TidesDB are done through transactions for ACID guarantees per column family.

Basic Transaction

/* Get column family pointer first */
tidesdb_column_family_t *cf = tidesdb_get_column_family(db, "my_cf");
if (!cf) return -1;

tidesdb_txn_t *txn = NULL;
if (tidesdb_txn_begin(db, &txn) != 0)
{
    return -1;
}

const uint8_t *key = (uint8_t *)"mykey";
const uint8_t *value = (uint8_t *)"myvalue";

if (tidesdb_txn_put(txn, cf, key, 5, value, 7, -1) != 0)
{
    tidesdb_txn_free(txn);
    return -1;
}

if (tidesdb_txn_commit(txn) != 0)
{
    tidesdb_txn_free(txn);
    return -1;
}

tidesdb_txn_free(txn);

With TTL (Time-to-Live)

tidesdb_column_family_t *cf = tidesdb_get_column_family(db, "my_cf");
if (!cf) return -1;

tidesdb_txn_t *txn = NULL;
tidesdb_txn_begin(db, &txn);

const uint8_t *key = (uint8_t *)"temp_key";
const uint8_t *value = (uint8_t *)"temp_value";

/* TTL is Unix timestamp (seconds since epoch) -- absolute expiration time */
time_t ttl = time(NULL) + 60;  /* Expires 60 seconds from now */

/* Use -1 for no expiration */
tidesdb_txn_put(txn, cf, key, 8, value, 10, ttl);
tidesdb_txn_commit(txn);
tidesdb_txn_free(txn);

TTL Examples

time_t ttl = -1;

time_t ttl = time(NULL) + (5 * 60);

time_t ttl = time(NULL) + (60 * 60);

time_t ttl = 1730592000;

Getting a Key-Value Pair

tidesdb_column_family_t *cf = tidesdb_get_column_family(db, "my_cf");
if (!cf) return -1;

tidesdb_txn_t *txn = NULL;
tidesdb_txn_begin(db, &txn);

const uint8_t *key = (uint8_t *)"mykey";
uint8_t *value = NULL;
size_t value_size = 0;

if (tidesdb_txn_get(txn, cf, key, 5, &value, &value_size) == 0)
{
    /* Use value */
    printf("Value: %.*s\n", (int)value_size, value);
    free(value);
}

tidesdb_txn_free(txn);

Deleting a Key-Value Pair

tidesdb_column_family_t *cf = tidesdb_get_column_family(db, "my_cf");
if (!cf) return -1;

tidesdb_txn_t *txn = NULL;
tidesdb_txn_begin(db, &txn);

const uint8_t *key = (uint8_t *)"mykey";
tidesdb_txn_delete(txn, cf, key, 5);

tidesdb_txn_commit(txn);
tidesdb_txn_free(txn);

Multi-Operation Transaction

tidesdb_column_family_t *cf = tidesdb_get_column_family(db, "my_cf");
if (!cf) return -1;

tidesdb_txn_t *txn = NULL;
tidesdb_txn_begin(db, &txn);

tidesdb_txn_put(txn, cf, (uint8_t *)"key1", 4, (uint8_t *)"value1", 6, -1);
tidesdb_txn_put(txn, cf, (uint8_t *)"key2", 4, (uint8_t *)"value2", 6, -1);
tidesdb_txn_delete(txn, cf, (uint8_t *)"old_key", 7);

/* Commit atomically -- all or nothing */
if (tidesdb_txn_commit(txn) != 0)
{
    tidesdb_txn_free(txn);
    return -1;
}

tidesdb_txn_free(txn);

Transaction Rollback

tidesdb_column_family_t *cf = tidesdb_get_column_family(db, "my_cf");
if (!cf) return -1;

tidesdb_txn_t *txn = NULL;
tidesdb_txn_begin(db, &txn);

tidesdb_txn_put(txn, cf, (uint8_t *)"key", 3, (uint8_t *)"value", 5, -1);

tidesdb_txn_rollback(txn);
tidesdb_txn_free(txn);

Transaction Reset

tidesdb_txn_reset resets a committed or aborted transaction for reuse with a new isolation level. This avoids the overhead of freeing and reallocating transaction resources in hot loops.

int tidesdb_txn_reset(tidesdb_txn_t *txn, tidesdb_isolation_level_t isolation);

Parameters

Name	Type	Description
txn	tidesdb_txn_t*	Transaction handle (must be committed or aborted)
isolation	tidesdb_isolation_level_t	New isolation level for the reset transaction

Returns

• TDB_SUCCESS on success
• TDB_ERR_INVALID_ARGS if txn is NULL, still active (not committed/aborted), or isolation level is invalid

Example

tidesdb_column_family_t *cf = tidesdb_get_column_family(db, "my_cf");
if (!cf) return -1;

tidesdb_txn_t *txn = NULL;
tidesdb_txn_begin(db, &txn);

tidesdb_txn_put(txn, cf, (uint8_t *)"key1", 4, (uint8_t *)"value1", 6, -1);
tidesdb_txn_commit(txn);

tidesdb_txn_reset(txn, TDB_ISOLATION_READ_COMMITTED);

tidesdb_txn_put(txn, cf, (uint8_t *)"key2", 4, (uint8_t *)"value2", 6, -1);
tidesdb_txn_commit(txn);

tidesdb_txn_free(txn);

Behavior

• The transaction must be committed or aborted before reset; resetting an active transaction returns TDB_ERR_INVALID_ARGS
• Internal buffers (ops array, read set arrays, arena pointer array, column family array, savepoints array) are retained to avoid reallocation
• Per-operation key/value data, arena buffers, hash tables, and savepoint children are freed
• A fresh txn_id and snapshot_seq are assigned based on the new isolation level
• The isolation level can be changed on each reset (e.g., READ_COMMITTED → REPEATABLE_READ)
• If switching to an isolation level that requires read tracking (REPEATABLE_READ or higher), read set arrays are allocated automatically
• SERIALIZABLE transactions are correctly unregistered from and re-registered to the active transaction list

When to use

• Batch processing · Reuse a single transaction across many commit cycles in a loop
• Connection pooling · Reset a transaction for a new request without reallocation
• High-throughput ingestion · Reduce malloc/free overhead in tight write loops

Reset vs Free + Begin

For a single transaction, tidesdb_txn_reset is functionally equivalent to calling tidesdb_txn_free followed by tidesdb_txn_begin_with_isolation. The difference is performance: reset retains allocated buffers and avoids repeated allocation overhead. This matters most in loops that commit and restart thousands of transactions.

Savepoints

Savepoints allow partial rollback within a transaction. You can create named savepoints and rollback to them without aborting the entire transaction.

tidesdb_column_family_t *cf = tidesdb_get_column_family(db, "my_cf");
if (!cf) return -1;

tidesdb_txn_t *txn = NULL;
tidesdb_txn_begin(db, &txn);

tidesdb_txn_put(txn, cf, (uint8_t *)"key1", 4, (uint8_t *)"value1", 6, -1);

if (tidesdb_txn_savepoint(txn, "sp1") != 0)
{
    tidesdb_txn_rollback(txn);
    tidesdb_txn_free(txn);
    return -1;
}

tidesdb_txn_put(txn, cf, (uint8_t *)"key2", 4, (uint8_t *)"value2", 6, -1);

if (tidesdb_txn_rollback_to_savepoint(txn, "sp1") != 0)
{
    tidesdb_txn_rollback(txn);
    tidesdb_txn_free(txn);
    return -1;
}

tidesdb_txn_put(txn, cf, (uint8_t *)"key3", 4, (uint8_t *)"value3", 6, -1);

if (tidesdb_txn_commit(txn) != 0)
{
    tidesdb_txn_free(txn);
    return -1;
}

tidesdb_txn_free(txn);

Savepoint API

• tidesdb_txn_savepoint(txn, "name") · Create a savepoint
• tidesdb_txn_rollback_to_savepoint(txn, "name") · Rollback to savepoint
• tidesdb_txn_release_savepoint(txn, "name") · Release savepoint without rolling back

Savepoint behavior

• Savepoints capture the transaction state at a specific point
• Rolling back to a savepoint discards all operations after that savepoint
• Releasing a savepoint frees its resources without rolling back
• Multiple savepoints can be created with different names
• Creating a savepoint with an existing name updates that savepoint
• Savepoints are automatically freed when the transaction commits or rolls back
• Returns TDB_ERR_NOT_FOUND if the savepoint name doesn’t exist

Multi-Column-Family Transactions

TidesDB supports atomic transactions across multiple column families with true all-or-nothing semantics.

tidesdb_column_family_t *users_cf = tidesdb_get_column_family(db, "users");
tidesdb_column_family_t *orders_cf = tidesdb_get_column_family(db, "orders");
if (!users_cf || !orders_cf) return -1;

tidesdb_txn_t *txn = NULL;
if (tidesdb_txn_begin(db, &txn) != 0)
{
    return -1;
}

tidesdb_txn_put(txn, users_cf, (uint8_t *)"user:1000", 9,
                (uint8_t *)"John Doe", 8, -1);

tidesdb_txn_put(txn, orders_cf, (uint8_t *)"order:5000", 10,
                (uint8_t *)"user:1000|product:A", 19, -1);

if (tidesdb_txn_commit(txn) != 0)
{
    tidesdb_txn_free(txn);
    return -1;
}

tidesdb_txn_free(txn);

Multi-CF guarantees

• Either all CFs commit or none do (atomic)
• Automatically detected when operations span multiple CFs
• Uses global sequence numbers for atomic ordering
• Each CF’s WAL receives operations with the same commit sequence number
• No two-phase commit or coordinator overhead

Isolation Levels

TidesDB supports five MVCC isolation levels for fine-grained concurrency control.

tidesdb_column_family_t *cf = tidesdb_get_column_family(db, "my_cf");
if (!cf) return -1;

tidesdb_txn_t *txn = NULL;

tidesdb_txn_begin_with_isolation(db, TDB_ISOLATION_READ_UNCOMMITTED, &txn);
tidesdb_txn_begin_with_isolation(db, TDB_ISOLATION_READ_COMMITTED, &txn);
tidesdb_txn_begin_with_isolation(db, TDB_ISOLATION_REPEATABLE_READ, &txn);
tidesdb_txn_begin_with_isolation(db, TDB_ISOLATION_SNAPSHOT, &txn);
tidesdb_txn_begin_with_isolation(db, TDB_ISOLATION_SERIALIZABLE, &txn);

tidesdb_txn_put(txn, cf, (uint8_t *)"key", 3, (uint8_t *)"value", 5, -1);

int result = tidesdb_txn_commit(txn);
if (result == TDB_ERR_CONFLICT)
{
    tidesdb_txn_free(txn);
    return -1;
}

tidesdb_txn_free(txn);

Isolation level characteristics

• READ UNCOMMITTED · Maximum concurrency, minimal consistency
• READ COMMITTED · Balanced for OLTP workloads (default)
• REPEATABLE READ · Strong point read consistency
• SNAPSHOT · Prevents lost updates with write-write conflict detection
• SERIALIZABLE · Strongest guarantees with full SSI, higher abort rates

Iterators

Iterators provide efficient forward and backward traversal over key-value pairs.

Memory Ownership

The key and value pointers returned by tidesdb_iter_key() and tidesdb_iter_value() are internal pointers owned by the iterator. Do NOT free them. They remain valid until the next tidesdb_iter_next(), tidesdb_iter_prev(), seek operation, or tidesdb_iter_free() call. If you need to retain the data beyond the current iteration step, copy it to your own buffer.

Forward Iteration

tidesdb_column_family_t *cf = tidesdb_get_column_family(db, "my_cf");
if (!cf) return -1;

tidesdb_txn_t *txn = NULL;
tidesdb_txn_begin(db, &txn);

tidesdb_iter_t *iter = NULL;
if (tidesdb_iter_new(txn, cf, &iter) != 0)
{
    tidesdb_txn_free(txn);
    return -1;
}

tidesdb_iter_seek_to_first(iter);

while (tidesdb_iter_valid(iter))
{
    uint8_t *key = NULL;
    size_t key_size = 0;
    uint8_t *value = NULL;
    size_t value_size = 0;

    if (tidesdb_iter_key(iter, &key, &key_size) == 0 &&
        tidesdb_iter_value(iter, &value, &value_size) == 0)
    {
        printf("Key: %.*s, Value: %.*s\n",
               (int)key_size, key, (int)value_size, value);
    }

    tidesdb_iter_next(iter);
}

tidesdb_iter_free(iter);
tidesdb_txn_free(txn);

Backward Iteration

tidesdb_column_family_t *cf = tidesdb_get_column_family(db, "my_cf");
if (!cf) return -1;

tidesdb_txn_t *txn = NULL;
tidesdb_txn_begin(db, &txn);

tidesdb_iter_t *iter = NULL;
tidesdb_iter_new(txn, cf, &iter);

tidesdb_iter_seek_to_last(iter);

while (tidesdb_iter_valid(iter))
{
    tidesdb_iter_prev(iter);
}

tidesdb_iter_free(iter);
tidesdb_txn_free(txn);

Iterator Seek Operations

TidesDB provides seek operations that allow you to position an iterator at a specific key or key range without scanning from the beginning.

How Seek Works

With Block Indexes Enabled (enable_block_indexes = 1):

• Uses compact block index with parallel arrays (min/max key prefixes and file positions)
• Binary search through sampled keys at configurable ratio (default 1:1 via index_sample_ratio, meaning every block is indexed)
• Jumps directly to the target block using the file position
• Scans forward from that block to find the exact key
• Performance · O(log n) binary search + O(k) entries per block scan

Block indexes provide dramatic speedup for large SSTables at the cost of ~2-5% storage overhead for the compact index structure (parallel arrays with delta-encoded file positions).

Seek to Specific Key

tidesdb_iter_seek(iter, key, key_size) · Positions iterator at the first key >= target key

tidesdb_column_family_t *cf = tidesdb_get_column_family(db, "my_cf");
if (!cf) return -1;

tidesdb_txn_t *txn = NULL;
tidesdb_txn_begin(db, &txn);

tidesdb_iter_t *iter = NULL;
tidesdb_iter_new(txn, cf, &iter);

/* Seek to specific key */
const char *target = "user:1000";
if (tidesdb_iter_seek(iter, (uint8_t *)target, strlen(target)) == 0)
{
    /* Iterator is now positioned at "user:1000" or the next key after it */
    if (tidesdb_iter_valid(iter))
    {
        uint8_t *key = NULL;
        size_t key_size = 0;
        tidesdb_iter_key(iter, &key, &key_size);
        printf("Found: %.*s\n", (int)key_size, key);
    }
}

tidesdb_iter_free(iter);
tidesdb_txn_free(txn);

tidesdb_iter_seek_for_prev(iter, key, key_size) · Positions iterator at the last key <= target key

/* Seek for reverse iteration */
const char *target = "user:2000";
if (tidesdb_iter_seek_for_prev(iter, (uint8_t *)target, strlen(target)) == 0)
{
    /* Iterator is now positioned at "user:2000" or the previous key before it */
    while (tidesdb_iter_valid(iter))
    {
        /* Iterate backwards from this point */
        tidesdb_iter_prev(iter);
    }
}

Prefix Seeking

Since tidesdb_iter_seek positions the iterator at the first key >= target, you can use a prefix as the seek target to efficiently scan all keys sharing that prefix:

/* Seek to prefix and iterate all matching keys */
const char *prefix = "user:";
if (tidesdb_iter_seek(iter, (uint8_t *)prefix, strlen(prefix) + 1) == 0)
{
    while (tidesdb_iter_valid(iter))
    {
        uint8_t *key = NULL;
        size_t key_size = 0;
        tidesdb_iter_key(iter, &key, &key_size);

        /* Stop when keys no longer match prefix */
        if (strncmp((char *)key, prefix, strlen(prefix)) != 0) break;

        /* Process key */
        printf("Found: %.*s\n", (int)key_size, key);

        if (tidesdb_iter_next(iter) != TDB_SUCCESS) break;
    }
}

This pattern works across both memtables and SSTables. When block indexes are enabled, the seek operation uses binary search to jump directly to the relevant block, making prefix scans efficient even on large datasets.

Custom Comparators

TidesDB uses comparators to determine the sort order of keys throughout the entire system: memtables, SSTables, block indexes, and iterators all use the same comparison logic. Once a comparator is set for a column family, it cannot be changed without corrupting data.

Built-in Comparators

TidesDB provides six built-in comparators that are automatically registered on database open:

"memcmp" (default) · Binary byte-by-byte comparison

• Compares min(key1_size, key2_size) bytes using memcmp()
• If bytes are equal, shorter key sorts first
• Use case · Binary keys, raw byte data, general purpose

"lexicographic" · Null-terminated string comparison

• Uses strcmp() for lexicographic ordering
• Ignores key_size parameters (assumes null-terminated)
• Use case · C strings, text keys
• Warning · Keys must be null-terminated or behavior is undefined

"uint64" · Unsigned 64-bit integer comparison

• Interprets 8-byte keys as uint64_t values
• Falls back to memcmp if key_size != 8
• Use case · Numeric IDs, timestamps, counters
• Example · uint64_t id = 1000; tidesdb_txn_put(txn, cf, (uint8_t*)&id, 8, ...)

"int64" · Signed 64-bit integer comparison

• Interprets 8-byte keys as int64_t values
• Falls back to memcmp if key_size != 8
• Use case · Signed numeric keys, relative timestamps
• Example · int64_t offset = -500; tidesdb_txn_put(txn, cf, (uint8_t*)&offset, 8, ...)

"reverse" · Reverse binary comparison

• Negates the result of memcmp comparator
• Sorts keys in descending order
• Use case · Reverse chronological order, descending IDs

"case_insensitive" · Case-insensitive ASCII comparison

• Converts A-Z to a-z during comparison
• Compares min(key1_size, key2_size) bytes
• If bytes are equal (ignoring case), shorter key sorts first
• Use case · Case-insensitive text keys, usernames, email addresses

Custom Comparator Registration

/* Define your comparison function */
int my_timestamp_compare(const uint8_t *key1, size_t key1_size,
                         const uint8_t *key2, size_t key2_size, void *ctx)
{
    (void)ctx;  /* unused */

    if (key1_size != 8 || key2_size != 8)
    {
        /* fallback for invalid sizes */
        return memcmp(key1, key2, key1_size < key2_size ? key1_size : key2_size);
    }

    uint64_t ts1, ts2;
    memcpy(&ts1, key1, 8);
    memcpy(&ts2, key2, 8);

    /* reverse order for newest-first */
    if (ts1 > ts2) return -1;
    if (ts1 < ts2) return 1;
    return 0;
}

/* Register before creating column families */
tidesdb_register_comparator(db, "timestamp_desc", my_timestamp_compare, NULL, NULL);

/* Use in column family */
tidesdb_column_family_config_t cf_config = tidesdb_default_column_family_config();
strncpy(cf_config.comparator_name, "timestamp_desc", TDB_MAX_COMPARATOR_NAME - 1);
cf_config.comparator_name[TDB_MAX_COMPARATOR_NAME - 1] = '\0';
tidesdb_create_column_family(db, "events", &cf_config);

Retrieving a Registered Comparator

Use tidesdb_get_comparator to retrieve a previously registered comparator by name:

tidesdb_comparator_fn fn = NULL;
void *ctx = NULL;

if (tidesdb_get_comparator(db, "timestamp_desc", &fn, &ctx) == 0)
{
    /* Comparator found - fn and ctx are now populated */
    printf("Comparator 'timestamp_desc' is registered\n");
}
else
{
    /* Comparator not found */
    printf("Comparator not registered\n");
}

Use cases

• Validation · Check if a comparator is registered before creating a column family
• Debugging · Verify comparator registration during development
• Dynamic configuration · Query available comparators at runtime

Comparator function signature

int (*comparator_fn)(const uint8_t *key1, size_t key1_size,
                     const uint8_t *key2, size_t key2_size,
                     void *ctx);

Return values

• < 0 if key1 < key2
• 0 if key1 == key2
• > 0 if key1 > key2

Important

Comparators must be registered before creating column families that use them. Once set, a comparator cannot be changed for a column family. The same comparator is used across memtables, SSTables, block indexes, and iterators.

• Custom comparators can use the ctx parameter for runtime configuration

Sync Modes

Control durability vs performance tradeoff with three sync modes.

tidesdb_column_family_config_t cf_config = tidesdb_default_column_family_config();

/* TDB_SYNC_NONE     -- Fastest, least durable (OS handles flushing) */
cf_config.sync_mode = TDB_SYNC_NONE;

/* TDB_SYNC_INTERVAL -- Balanced performance with periodic background syncing */
cf_config.sync_mode = TDB_SYNC_INTERVAL;
cf_config.sync_interval_us = 128000;  /* Sync every 128ms (default) */

/* TDB_SYNC_FULL     -- Most durable (fsync on every write) */
cf_config.sync_mode = TDB_SYNC_FULL;

tidesdb_create_column_family(db, "my_cf", &cf_config);

Sync Mode Options

• TDB_SYNC_NONE · No explicit sync, relies on OS page cache (fastest, least durable)

  • Best for · Maximum throughput, acceptable data loss on crash
  • Use case · Caches, temporary data, reproducible workloads

• TDB_SYNC_INTERVAL · Periodic background syncing at configurable intervals (balanced)

  • Best for · Production workloads requiring good performance with bounded data loss
  • Use case · Most applications, configurable durability window
  • Features:
    • Single background sync thread monitors all column families using interval mode
    • Configurable sync interval via sync_interval_us (microseconds)
    • Structural operations (flush, compaction, WAL rotation) always enforce durability
    • At most sync_interval_us worth of data at risk on crash
    • Mid-durability correctness - When memtables rotate, the WAL receives an escalated fsync before entering the flush queue. Sorted run creation and merge operations always propagate full sync to block managers regardless of sync mode.

• TDB_SYNC_FULL · Fsync on every write operation (slowest, most durable)

  • Best for · Critical data requiring maximum durability
  • Use case · Financial transactions, audit logs, critical metadata
  • Note: Structural operations always use fsync regardless of sync mode :::

Sync Interval Examples

/* Sync every 100ms (good for low-latency requirements) */
cf_config.sync_mode = TDB_SYNC_INTERVAL;
cf_config.sync_interval_us = 100000;

/* Sync every 128ms (default) */
cf_config.sync_mode = TDB_SYNC_INTERVAL;
cf_config.sync_interval_us = 128000;

/* Sync every 1 second (higher throughput, more data at risk) */
cf_config.sync_mode = TDB_SYNC_INTERVAL;
cf_config.sync_interval_us = 1000000;

Structural Operations

Regardless of sync mode, TidesDB always enforces durability for structural operations:

• Memtable flush to SSTable
• SSTable compaction and merging
• WAL rotation
• Column family metadata updates

This ensures the database structure remains consistent even if user data syncing is delayed.

Compaction

TidesDB performs automatic background compaction when L1 reaches the configured file count trigger (default: 4 SSTables). However, you can manually trigger compaction for specific scenarios.

Checking Flush/Compaction Status

Check if a column family currently has flush or compaction operations in progress.

tidesdb_column_family_t *cf = tidesdb_get_column_family(db, "my_cf");
if (!cf) return -1;

/* Check if flushing is in progress */
if (tidesdb_is_flushing(cf))
{
    printf("Flush in progress\n");
}

/* Check if compaction is in progress */
if (tidesdb_is_compacting(cf))
{
    printf("Compaction in progress\n");
}

Use cases

• Graceful shutdown · Wait for background operations to complete before closing
• Maintenance windows · Check if operations are running before triggering manual compaction
• Monitoring · Track background operation status for observability
• Testing · Verify flush/compaction behavior in unit tests

Return values

• 1 · Operation is in progress
• 0 · No operation in progress (or invalid column family)

Manual Flush

Manually flush a column family’s memtable to disk. This creates a new SSTable (sorted run) in level 1.

tidesdb_column_family_t *cf = tidesdb_get_column_family(db, "my_cf");
if (!cf) return -1;

/* Trigger flush manually */
if (tidesdb_flush_memtable(cf) != 0)
{
    fprintf(stderr, "Failed to trigger flush\n");
    return -1;
}

When to use manual flush

• Before backup · Ensure all in-memory data is persisted before taking a backup
• Memory pressure · Force data to disk when memory usage is high
• Testing · Verify SSTable creation and compaction behavior
• Graceful shutdown · Flush pending data before closing the database

Behavior

• Enqueues flush work in the global flush thread pool
• Returns immediately (non-blocking) — flush runs asynchronously in background threads
• If flush is already running for the column family, the call succeeds but doesn’t queue duplicate work
• Thread-safe — can be called concurrently from multiple threads

Manual Compaction

tidesdb_column_family_t *cf = tidesdb_get_column_family(db, "my_cf");
if (!cf) return -1;

/* Trigger compaction manually */
if (tidesdb_compact(cf) != 0)
{
    fprintf(stderr, "Failed to trigger compaction\n");
    return -1;
}

When to use manual compaction

• After bulk deletes · Reclaim disk space by removing tombstones and obsolete versions
• After bulk updates · Consolidate multiple versions of keys into single entries
• Before read-heavy workloads · Optimize read performance by reducing the number of levels to search
• During maintenance windows · Proactively compact during low-traffic periods to avoid compaction during peak load
• After TTL expiration · Remove expired entries to reclaim storage
• Space optimization · Force compaction to reduce space amplification when storage is constrained

Behavior

• Enqueues compaction work in the global compaction thread pool
• Returns immediately (non-blocking) — compaction runs asynchronously in background threads
• If compaction is already running for the column family, the call succeeds but doesn’t queue duplicate work
• Compaction merges SSTables across levels, removes tombstones, expired TTL entries, and obsolete versions
• Thread-safe — can be called concurrently from multiple threads

Performance considerations

• Manual compaction uses the same thread pool as automatic background compaction
• Configure thread pool size via config.num_compaction_threads (default: 2)
• Compaction is I/O intensive — avoid triggering during peak write workloads
• Multiple column families can compact in parallel up to the thread pool limit

See How does TidesDB work? for details on compaction algorithms, merge strategies, and parallel compaction.

Thread Pools

TidesDB uses separate thread pools for flush and compaction operations. Understanding the parallelism model is important for optimal configuration.

Parallelism semantics

• Cross-CF parallelism · Multiple flush/compaction workers CAN process different column families in parallel
• Within-CF serialization · A single column family can only have one flush and one compaction running at any time (enforced by atomic is_flushing and is_compacting flags)
• No intra-CF memtable parallelism · Even if a CF has multiple immutable memtables queued, they are flushed sequentially

Thread pool sizing guidance

• Single column family · Set num_flush_threads = 1 and num_compaction_threads = 1. Additional threads provide no benefit since only one operation per CF can run at a time — extra threads will simply wait idle.
• Multiple column families · Set thread counts up to the number of column families for maximum parallelism. With N column families and M workers (where M ≤ N), throughput scales linearly.

Configuration

tidesdb_config_t config = {
    .db_path = "./mydb",
    .num_flush_threads = 2,                /* Flush thread pool size (default: 2) */
    .num_compaction_threads = 2,           /* Compaction thread pool size (default: 2) */
    .log_level = TDB_LOG_INFO,
    .block_cache_size = 64 * 1024 * 1024,  /* 64MB global block cache (default: 64MB) */
    .max_open_sstables = 256,              /* LRU cache for SSTable objects (default: 256, each has 2 FDs) */
};

tidesdb_t *db = NULL;
tidesdb_open(&config, &db);

Note

max_open_sstables is a storage-engine-level configuration, not a column family configuration. It controls the LRU cache size for SSTable structures. Each SSTable uses 2 file descriptors (klog + vlog), so 256 SSTables = 512 file descriptors.

See How does TidesDB work? for details on thread pool architecture and work distribution.

Utility Functions

tidesdb_free

Use tidesdb_free to free memory allocated by TidesDB. This is particularly useful for FFI/language bindings where the caller needs to free memory returned by TidesDB functions (e.g., values from tidesdb_txn_get).

uint8_t *value = NULL;
size_t value_size = 0;

if (tidesdb_txn_get(txn, cf, key, key_size, &value, &value_size) == 0)
{
    /* Use value */
    printf("Value: %.*s\n", (int)value_size, value);

    /* Free using tidesdb_free (or standard free) */
    tidesdb_free(value);
}

When to use tidesdb_free

• FFI bindings · Language bindings (Java, Rust, Go, Python) should use tidesdb_free to ensure memory is freed by the same allocator that allocated it
• Cross-platform · Ensures correct deallocation on all platforms
• Consistency · Provides a uniform API for memory management

For native C/C++ applications, free() works identically since TidesDB uses the standard allocator.