15 721
play

15-721 DATABASE SYSTEMS Lecture #06 Index Locking & Latching - PowerPoint PPT Presentation

Jul 25, 2023 •438 likes •1.49k views

15-721 DATABASE SYSTEMS Lecture #06 Index Locking & Latching Andy Pavlo / / Carnegie Mellon University / / Spring 2016 @Andy_Pavlo // Carnegie Mellon University // Spring 2017 2 TODAYS AGENDA Index Locks vs. Latches Latch

  1. 18 LATCH CRABBING Search: Start at root and go down; repeatedly, → Acquire read ( R ) latch on child → Then unlock parent if the child is safe. Insert/Delete: Start at root and go down, obtaining write ( W ) latches as needed. Once child is locked, check if it is safe: → If child is safe, release all locks on ancestors. CMU 15-721 (Spring 2017)

  2. 19 EXAMPLE #1: SEARCH 23 A 20 B C 10 35 D E F G 6 12 23 38 44 CMU 15-721 (Spring 2017)

  3. 19 EXAMPLE #1: SEARCH 23 R A 20 B C 10 35 D E F G 6 12 23 38 44 CMU 15-721 (Spring 2017)

  4. 19 EXAMPLE #1: SEARCH 23 We can release the latch on A as R soon as we acquire the latch for C . A 20 R B C 10 35 D E F G 6 12 23 38 44 CMU 15-721 (Spring 2017)

  5. 19 EXAMPLE #1: SEARCH 23 We can release the latch on A as soon as we acquire the latch for C . A 20 R B C 10 35 D E F G 6 12 23 38 44 CMU 15-721 (Spring 2017)

  6. 19 EXAMPLE #1: SEARCH 23 We can release the latch on A as soon as we acquire the latch for C . A 20 R B C 10 35 R D E F G 6 12 23 38 44 CMU 15-721 (Spring 2017)

  7. 19 EXAMPLE #1: SEARCH 23 We can release the latch on A as soon as we acquire the latch for C . A 20 B C 10 35 R D E F G 6 12 23 38 44 CMU 15-721 (Spring 2017)

  8. 20 EXAMPLE #2: DELETE 44 A 20 B C 10 35 D E F G 6 12 23 38 44 CMU 15-721 (Spring 2017)

  9. 20 EXAMPLE #2: DELETE 44 W A 20 B C 10 35 D E F G 6 12 23 38 44 CMU 15-721 (Spring 2017)

  10. 20 EXAMPLE #2: DELETE 44 We may need to coalesce C , so we W can’t release the latch on A . A 20 W B C 10 35 D E F G 6 12 23 38 44 CMU 15-721 (Spring 2017)

  11. 20 EXAMPLE #2: DELETE 44 We may need to coalesce C , so we W can’t release the latch on A . A 20 G will not merge with F , so we can release latches on A and C . W B C 10 35 W D E F G 6 12 23 38 44 CMU 15-721 (Spring 2017)

  12. 20 EXAMPLE #2: DELETE 44 We may need to coalesce C , so we can’t release the latch on A . A 20 G will not merge with F , so we can release latches on A and C . B C 10 35 X W D E F G 6 12 23 38 44 CMU 15-721 (Spring 2017)

  13. 21 EXAMPLE #3: INSERT 40 A 20 B C 10 35 D E F G 6 12 23 38 44 CMU 15-721 (Spring 2017)

  14. 21 EXAMPLE #3: INSERT 40 W A 20 B C 10 35 D E F G 6 12 23 38 44 CMU 15-721 (Spring 2017)

  15. 21 EXAMPLE #3: INSERT 40 C has room if its child has to split, W so we can release the latch on A . A 20 W B C 10 35 D E F G 6 12 23 38 44 CMU 15-721 (Spring 2017)

  16. 21 EXAMPLE #3: INSERT 40 C has room if its child has to split, so we can release the latch on A . A 20 W B C 10 35 D E F G 6 12 23 38 44 CMU 15-721 (Spring 2017)

  17. 21 EXAMPLE #3: INSERT 40 C has room if its child has to split, so we can release the latch on A . A 20 G has to split, so we can’t release the latch on C . W B C 10 35 W D E F G 6 12 23 38 44 CMU 15-721 (Spring 2017)

  18. 21 EXAMPLE #3: INSERT 40 C has room if its child has to split, so we can release the latch on A . A 20 G has to split, so we can’t release the latch on C . W B C 10 35 44 W D E F G H 6 12 23 38 44 40 44 CMU 15-721 (Spring 2017)

  19. 22 OBSERVATION What was the first step that the DBMS took in the two examples that updated the index? Delete 44 Insert 40 W W A A 20 20 CMU 15-721 (Spring 2017)

  20. 23 BETTER LATCH CRABBING Optimistically assume that the leaf is safe. → Take R latches as you traverse the tree to reach it and verify. → If leaf is not safe, then do previous algorithm. CONCURRENCY OF OPERATIONS ON B-TREES Acta Informatica 9: 1-21 1977 CMU 15-721 (Spring 2017)

  21. 24 EXAMPLE #4: DELETE 44 R A 20 B C 10 35 D E F G 6 12 23 38 44 CMU 15-721 (Spring 2017)

  22. 24 EXAMPLE #4: DELETE 44 We assume that C is safe, so we R can release the latch on A . A 20 R B C 10 35 D E F G 6 12 23 38 44 CMU 15-721 (Spring 2017)

  23. 24 EXAMPLE #4: DELETE 44 We assume that C is safe, so we can release the latch on A . A 20 R B C 10 35 D E F G 6 12 23 38 44 CMU 15-721 (Spring 2017)

  24. 24 EXAMPLE #4: DELETE 44 We assume that C is safe, so we can release the latch on A . A 20 Acquire an exclusive latch on G . R B C 10 35 D E F G 6 12 23 38 44 CMU 15-721 (Spring 2017)

  25. 24 EXAMPLE #4: DELETE 44 We assume that C is safe, so we can release the latch on A . A 20 Acquire an exclusive latch on G . B C 10 35 W D E F G 6 12 23 38 44 CMU 15-721 (Spring 2017)

  26. 24 EXAMPLE #4: DELETE 44 We assume that C is safe, so we can release the latch on A . A 20 Acquire an exclusive latch on G . B C 10 35 X W D E F G 6 12 23 38 44 CMU 15-721 (Spring 2017)

  27. 25 OBSERVATION Crabbing ensures that txns do not corrupt the internal data structure during modifications. But because txns release latches on each node as soon as they are finished their operations, we cannot guarantee that phantoms do not occur… CMU 15-721 (Spring 2017)

  28. 26 PROBLEM SCENARIO #1 A 20 B C 10 35 D E F G 6 12 23 38 44 CMU 15-721 (Spring 2017)

  29. 26 PROBLEM SCENARIO #1 Txn #1: Check if 25 exists R A 20 B C 10 35 D E F G 6 12 23 38 44 CMU 15-721 (Spring 2017)

  30. 26 PROBLEM SCENARIO #1 Txn #1: Check if 25 exists R A 20 R B C 10 35 D E F G 6 12 23 38 44 CMU 15-721 (Spring 2017)

  31. 26 PROBLEM SCENARIO #1 Txn #1: Check if 25 exists A 20 R B C 10 35 D E F G 6 12 23 38 44 CMU 15-721 (Spring 2017)

  32. 26 PROBLEM SCENARIO #1 Txn #1: Check if 25 exists A 20 B C 10 35 R D E F G ! 6 12 23 38 44 CMU 15-721 (Spring 2017)

  33. 26 PROBLEM SCENARIO #1 Txn #1: Check if 25 exists A Txn #2: Insert 25 20 B C 10 35 D E F G 6 12 23 38 44 CMU 15-721 (Spring 2017)

  34. 26 PROBLEM SCENARIO #1 Txn #1: Check if 25 exists A Txn #2: Insert 25 20 B C 10 35 W D E F G 6 12 23 38 44 CMU 15-721 (Spring 2017)

  35. 26 PROBLEM SCENARIO #1 Txn #1: Check if 25 exists A Txn #2: Insert 25 20 B C 10 35 W D E F G 6 12 23 25 38 44 CMU 15-721 (Spring 2017)

  36. 26 PROBLEM SCENARIO #1 Txn #1: Check if 25 exists A Txn #2: Insert 25 20 Txn #1: Insert 25 B C 10 35 W D E F G 6 12 23 25 38 44 CMU 15-721 (Spring 2017)

  37. 27 PROBLEM SCENARIO #2 Txn #1: Scan [12, 23] A 20 B C 10 35 D E F G 6 12 23 38 44 CMU 15-721 (Spring 2017)

  38. 27 PROBLEM SCENARIO #2 Txn #1: Scan [12, 23] A 20 B C 10 35 R D E F G 6 12 23 38 44 CMU 15-721 (Spring 2017)

  39. 27 PROBLEM SCENARIO #2 Txn #1: Scan [12, 23] A 20 B C 10 35 R R D E F G 6 12 23 38 44 CMU 15-721 (Spring 2017)

  40. 27 PROBLEM SCENARIO #2 Txn #1: Scan [12, 23] A Txn #2: Insert 21 20 B C 10 35 D E F G 6 12 23 38 44 CMU 15-721 (Spring 2017)

  41. 27 PROBLEM SCENARIO #2 Txn #1: Scan [12, 23] A Txn #2: Insert 21 20 W B C 10 35 W D E F G 6 12 23 38 44 CMU 15-721 (Spring 2017)

  42. 27 PROBLEM SCENARIO #2 Txn #1: Scan [12, 23] A Txn #2: Insert 21 20 W B C 10 35 W D E F G 6 12 23 21 23 38 44 CMU 15-721 (Spring 2017)

  43. 27 PROBLEM SCENARIO #2 Txn #1: Scan [12, 23] A Txn #2: Insert 21 20 B C 10 35 D E F G 6 12 21 23 23 38 44 CMU 15-721 (Spring 2017)

  44. 27 PROBLEM SCENARIO #2 Txn #1: Scan [12, 23] A Txn #2: Insert 21 20 Txn #1: Scan [12, 23] B C 10 35 D E F G 6 12 23 21 23 38 44 CMU 15-721 (Spring 2017)

  45. 27 PROBLEM SCENARIO #2 Txn #1: Scan [12, 23] A Txn #2: Insert 21 20 Txn #1: Scan [12, 23] B C 10 35 R R D E F G 6 12 23 21 23 38 44 CMU 15-721 (Spring 2017)

  46. 28 INDEX LOCKS Need a way to protect the index’s logical contents from other txns to avoid phantoms. Difference with index latches: → Locks are held for the entire duration of a txn. → Only acquired at the leaf nodes. → Not physically stored in index data structure. CMU 15-721 (Spring 2017)

  47. 29 INDEX LOCKS Lock Table txn1 txn2 txn3 S • • • X S txn3 txn2 txn4 S • • • S S txn4 txn6 txn5 S • • • IX X CMU 15-721 (Spring 2017)

  48. 29 INDEX LOCKS Lock Table txn1 txn2 txn3 S • • • X S txn3 txn2 txn4 S • • • S S txn4 txn6 txn5 S • • • IX X CMU 15-721 (Spring 2017)

  49. 30 INDEX LOCKING SCHEMES Predicate Locks Key-Value Locks Gap Locks Key-Range Locks Hierarchical Locking CMU 15-721 (Spring 2017)

  50. 31 PREDICATE LOCKS Proposed locking scheme from System R. → Shared lock on the predicate in a WHERE clause of a SELECT query. → Exclusive lock on the predicate in a WHERE clause of any UPDATE , INSERT , or DELETE query. Never implemented in any system. THE NOTIONS OF CONSISTENCY AND PREDICATE LOCKS IN A DATABASE SYSTEM CACM 1976 CMU 15-721 (Spring 2017)

  51. 32 PREDICATE LOCKS SELECT SUM (balance) INSERT INTO account FROM account (name, balance) WHERE name = ‘Biggie’ VALUES (‘Biggie’, 100); Records in Table ‘account’ name=‘Biggie’ CMU 15-721 (Spring 2017)

  52. 32 PREDICATE LOCKS SELECT SUM (balance) INSERT INTO account FROM account (name, balance) WHERE name = ‘Biggie’ VALUES (‘Biggie’, 100); Records in Table ‘account’ name=‘Biggie’ name=‘Biggie’ ∧ balance=100 CMU 15-721 (Spring 2017)

  53. 33 KEY-VALUE LOCKS Locks that cover a single key value. Need “virtual keys” for non-existent values. B+Tree Leaf Node 10 12 14 16 CMU 15-721 (Spring 2017)

  54. 33 KEY-VALUE LOCKS Locks that cover a single key value. Need “virtual keys” for non-existent values. B+Tree Leaf Node Key [14, 14] 10 12 14 16 CMU 15-721 (Spring 2017)

  55. 34 GAP LOCKS Each txn acquires a key-value lock on the single key that it wants to access. Then get a gap lock on the next key gap. B+Tree Leaf Node 10 12 14 16 CMU 15-721 (Spring 2017)

  56. 34 GAP LOCKS Each txn acquires a key-value lock on the single key that it wants to access. Then get a gap lock on the next key gap. B+Tree Leaf Node 10 {Gap} 12 {Gap} 14 {Gap} 16 CMU 15-721 (Spring 2017)

  57. 34 GAP LOCKS Each txn acquires a key-value lock on the single key that it wants to access. Then get a gap lock on the next key gap. B+Tree Leaf Node 10 {Gap} 12 {Gap} 14 {Gap} 16 Gap (14, 16) CMU 15-721 (Spring 2017)

  58. 35 KEY-RANGE LOCKS A txn takes locks on ranges in the key space. → Each range is from one key that appears in the relation, to the next that appears. → Define lock modes so conflict table will capture commutativity of the operations available. CMU 15-721 (Spring 2017)

  59. 36 KEY-RANGE LOCKS Locks that cover a key value and the gap to the next key value in a single index. → Need “virtual keys” for artificial values (infinity) B+Tree Leaf Node 10 {Gap} 12 {Gap} 14 {Gap} 16 CMU 15-721 (Spring 2017)

  60. 36 KEY-RANGE LOCKS Locks that cover a key value and the gap to the next key value in a single index. → Need “virtual keys” for artificial values (infinity) B+Tree Leaf Node Next Key [14, 16) 10 {Gap} 12 {Gap} 14 {Gap} 16 CMU 15-721 (Spring 2017)

  61. 36 KEY-RANGE LOCKS Locks that cover a key value and the gap to the next key value in a single index. → Need “virtual keys” for artificial values (infinity) B+Tree Leaf Node 10 {Gap} 12 {Gap} 14 {Gap} 16 Prior Key (12, 14] CMU 15-721 (Spring 2017)

  62. 37 HIERARCHICAL LOCKING Allow for a txn to hold wider key-range locks with different locking modes. → Reduces the number of visits to lock manager. B+Tree Leaf Node 10 {Gap} 12 {Gap} 14 {Gap} 16 CMU 15-721 (Spring 2017)

  63. 37 HIERARCHICAL LOCKING Allow for a txn to hold wider key-range locks with different locking modes. → Reduces the number of visits to lock manager. IX B+Tree Leaf Node 10 {Gap} 12 {Gap} 14 {Gap} 16 [10, 16) CMU 15-721 (Spring 2017)

  64. 37 HIERARCHICAL LOCKING Allow for a txn to hold wider key-range locks with different locking modes. → Reduces the number of visits to lock manager. IX B+Tree Leaf Node X [14, 16) 10 {Gap} 12 {Gap} 14 {Gap} 16 [10, 16) CMU 15-721 (Spring 2017)

Recommend

More recommend