You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Copy file name to clipboardExpand all lines: src/content/docs/getting-started/what-is-tidesdb.md
+1-1Lines changed: 1 addition & 1 deletion
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -27,7 +27,7 @@ reads through bloom filters, caching, block indices, and compaction.
27
27
- Lock-free block manager using `pread`/`pwrite` for concurrent I/O. Reference-counted blocks with atomic operations. xxHash32 checksums for integrity. Supports up to 4GB blocks with partial reads.
28
28
- Two-tier caching
29
29
- File handle cache with LRU eviction (default 512 open SSTables). Background reaper closes oldest unused files.
30
-
-Block cache using partitioned CLOCK eviction. Caches deserialized klog blocks with zero-copy API and reference bit protection.
30
+
-NUMA-aware block cache using partitioned CLOCK eviction. Caches deserialized klog blocks with zero-copy API and reference bit protection.
31
31
- Background thread pools for flush and compaction (default 2 threads each). Work queues distribute tasks. Compaction auto-triggers when Level 1 reaches (default 4) files.
32
32
- Three sync modes · `TDB_SYNC_NONE` (OS-managed), `TDB_SYNC_FULL` (fsync every write), `TDB_SYNC_INTERVAL` (periodic sync). Structural operations always enforce durability.
33
33
- Compression support · LZ4,LZ4-FAST, Zstd, Snappy (configurable per column family). Applied to klog and vlog blocks, not WAL.
0 commit comments