1. Improvement WT-18781
    Implement fast-truncate at subtree level
  2. Task WT-18765
    History store dirty content accumulates unevictably during the checkpoint history store pass
  3. Task WT-18565
    Allow the cursor walk to skip clean in-memory internal pages
  4. Task WT-18564
    Extend the checkpoint-cleanup page-in dirty suppression to non-disaggregated btrees
  5. Task WT-18558
    [Disagg] Disable key-level garbage collection in checkpoint cleanup before private preview
  6. Improvement WT-18557
    Use LIST instead of TAILQ for remaining small hash tables (keyed encryptors, bucket storage, backup targets, background compact tracking, generic hash map)
  7. Improvement WT-18556
    Use LIST instead of TAILQ for the on-disk block cache hash table (WT_BLKCACHE)
  8. Improvement WT-18555
    Use LIST instead of TAILQ for the block manager and file handle hash tables (conn->blockhash, conn->fhhash)
  9. Improvement WT-18554
    Use LIST instead of TAILQ for the data handle hash tables (conn->dhhash, session->dhhash)
  10. Task WT-18550
    Checkpoint cleanup dirties a fully deleted disaggregated internal page for a no-progress rewrite before its truncate is globally visible
  11. Task WT-18525
    Evaluate skipping emptied internal pages for the local block manager
  12. Task WT-18524
    Evaluate skipping emptied internal pages on non-cursor walk and descent paths
  13. Bug WT-18496
    Reconciliation should merge the effective (page_del-applied) aggregate for WT_CELL_ADDR_DEL children into the parent, not the stale raw child aggregate
  14. Bug WT-18495
    Fix WT_CELL_ADDR_DEL time aggregate verification to not compare against stale child aggregate
  15. Task WT-18469
    Investigate the trade-off of letting eviction reconcile ahead of a running checkpoint in disaggregated storage
  16. Improvement WT-18427
    Non-forced drop discards the tree from cache synchronously while holding the schema and dhandle-list write locks
  17. Task WT-18407
    Prioritise eviction of internal pages whose children are fast-truncated
  18. Task WT-18406
    Skip reading an internal page when its time aggregate shows the whole subtree is deleted and visible
  19. Improvement WT-18388
    Reduce scratch buffer locking mechanism to only default session
  20. Improvement WT-18275
    Avoid per-key heap allocation when reconstructing leaf pages from deltas
  21. Improvement WT-17789
    Identify freeable updates to help eviction
  22. Improvement WT-17788
    Use list to replace tailq for cross checkpoint caching table
  23. Task WT-17736
    Avoid redundant layered-level lookup for insert/update in leader mode
  24. Task WT-17731
    Eliminate 7-byte alignment padding in __wt_page_modify caused by bool instantiated
  25. Improvement WT-17684
    Select older non-aborted update instead of NULL when skipping committed preserved prepared update during reconciliation
  26. Improvement WT-17681
    Skip reconciliation for pages already processed by eviction under the same checkpoint snapshot
  27. Improvement WT-17680
    Use checkpoint snapshot for eviction when precise checkpoint is enabled
  28. Task WT-17677
    Avoid marking btree and internal page dirty when instantiating updates on page read
  29. Task WT-17615
    Optimize __rts_btree_row_modify and __rts_btree_col_modify to skip redundant B-tree search
  30. Task WT-17601
    Do not initiate tombstones when reading a page into memory for disaggregated storage
  31. Task WT-17388
    Replace WT_ACQUIRE_BARRIER with explicit atomic loads in WT_TIME_WINDOW_SET_START/STOP macros
  32. Improvement WT-17050
    Avoid doing the search to check existing keys for layered insert and update on leader
  33. Improvement WT-17049
    Avoid reopening the stable table for each operation on leader
  34. Improvement WT-17030
    Avoid reopening the stable table for each operation on follower
  35. Task WT-16838
    Add software prefetch pipeline to WiredTiger B-tree search and cursor reopen path
  36. Improvement WT-16410
    Remove image size from block meta
  37. Improvement WT-16239
    Write a full page rather than a delta when a page has many globally visible deletes
  38. Task WT-16102
    Skip stable table lookup for insert and update operations on the follower node
  39. Task WT-15754
    Consider using relaxed atomic RMW and CAS for stats
  40. Improvement WT-12857
    Check the debug flags only in diagnostic build
  41. Improvement WT-12856
    Avoid null checking collator each time when compare keys
  42. Improvement WT-12855
    Track the depth of the btree when we split instead when we do search each time
  43. Improvement WT-12828
    Provide optimized versions of get/set_key/value for MongoDB record store use cases
  44. Improvement WT-12818
    No need to save the row key if the we already have a commit timestamp set
  45. Improvement WT-12811
    keep cached cursors buffers for a while
  46. Improvement WT-12803
    Optimise where we can skip copying keys for prepared transaction resolution
  47. Improvement WT-12798
    Checkpoint time is high when the cache size is large
  48. Task WT-12794
    Optimize search_near's cursor walk based on set bounds
  49. Improvement WT-12766
    Speed up WiredTiger implicit transactions
  50. Improvement WT-12725
    Perf improvement for allocation of cursor keys
Refresh results