Optimistic Algorithm and Concurrency Control Algorithm
12 likes4,591 views
The document discusses optimistic algorithms and concurrency control algorithms in operating systems, focusing on methods like the Kung-Robinson algorithm, which operates under the assumption that conflicts are rare and thus avoids synchronization during transaction execution. It outlines various centralized and distributed concurrency control techniques, including centralized locking and two-phase locking protocols, which manage locks to ensure consistency and handle transaction conflicts effectively. The document provides an overview of phases in these algorithms, emphasizing the efficiency of optimistic approaches and the mechanisms for detecting and resolving conflicts.
![The Algorithm (Kung-Robinson)
•Validation Phase (can be described as)
–T: a transaction
–ts: the highest sequence number at the start of T
–tf: the highest sequence number at the beginning of its validation phase
14
Shounak Katyayan Advances in Operating System Design
M. Tech CSE : 2014 - 15
valid:=true;
for t:=ts+1 to tf do
if (writeset(t) & readset[T] != {} ) then
valid :=false;
if valid then {write phase; increment counter, assign T a sequence number}](https://image.slidesharecdn.com/13optimisticalgorithmandconcurrencycontrolalgorithmshounakkatyayan-141127131505-conversion-gate01/85/Optimistic-Algorithm-and-Concurrency-Control-Algorithm-14-320.jpg)
Optimistic Algorithm and Concurrency Control Algorithm
- 1. Optimistic Algorithms & Concurrency Control Algorithms 1 Shounak Katyayan Advances in Operating System Design Shounak Katyayan M. Tech CSE : 2014 - 15
- 2. INDEX •Optimistic Algorithms –Kung Robinson Algorithm •Concurrency Control Algorithms –Completely Centralized Algorithm –Centralized Locking Algorithm –INGRES’ Primary-Site Locking Algorithm –Two-Phase Locking Algorithm M. Tech CSE : 2014 - 15 Shounak Katyayan Advances in Operating System Design 2
- 3. Optimistic Algorithms •Based on the assumption “conflicts do not occur during execution time” 3 Shounak Katyayan Advances in Operating System Design M. Tech CSE : 2014 - 15
- 4. Optimistic Algorithms •Thus, No synchronization is performed when a transaction is executed. 4 Shounak Katyayan Advances in Operating System Design M. Tech CSE : 2014 - 15
- 5. Optimistic Algorithms •However, A check is performed at the end of the transaction (to ensure no conflicts have occurred) 5 Shounak Katyayan Advances in Operating System Design M. Tech CSE : 2014 - 15
- 6. Optimistic Algorithms •If conflict == true –Transaction aborted •Else –Commit Transaction 6 Shounak Katyayan Advances in Operating System Design M. Tech CSE : 2014 - 15
- 7. Optimistic Algorithms •Since conflicts do not occur very often, this algorithm is very efficient compared to other locking algorithms. 7 Shounak Katyayan Advances in Operating System Design M. Tech CSE : 2014 - 15
- 8. Kung-Robinson Algorithm •H. T. KUNG and JOHN T. ROBINSON were the first to propose an optimistic method for concurrency control. •The optimistic situation for this algorithm happens when conflicts are unlikely to happen; the system consists mainly read-only transactions (such as a query dominant (powerful) system) •Basic Idea: No synchronization check is performed during transaction processing time, however, a validation is performed to make sure that no conflicts occurred. If a conflict is found, the tentative write is discarded and the transaction is restarted. 8 Shounak Katyayan Advances in Operating System Design M. Tech CSE : 2014 - 15
- 9. The Algorithm (Kung-Robinson) •Divided into three phases : Read Phase Validation Phase Write Phase 9 Shounak Katyayan Advances in Operating System Design M. Tech CSE : 2014 - 15
- 10. The Algorithm (Kung-Robinson) •Divided into three phases : Read Phase Validation Phase Write Phase 10 Shounak Katyayan Advances in Operating System Design M. Tech CSE : 2014 - 15 data objects are read, the intended computation of the transaction is done, and writes are made on a temporary storage.
- 11. The Algorithm (Kung-Robinson) •Divided into three phases : Read Phase Validation Phase Write Phase 11 Shounak Katyayan Advances in Operating System Design M. Tech CSE : 2014 - 15 check to see if writes made by the transaction violate the consistency of the database.
- 12. The Algorithm (Kung-Robinson) •Divided into three phases : Read Phase Validation Phase Write Phase 12 Shounak Katyayan Advances in Operating System Design M. Tech CSE : 2014 - 15 If the check finds out any conflicts, the data in the temporary storage will be discarded. Otherwise, the write phase will write the data into the database.
- 13. The Algorithm (Kung-Robinson) •Divided into three phases : Read Phase Validation Phase Write Phase 13 Shounak Katyayan Advances in Operating System Design M. Tech CSE : 2014 - 15 If the validation phase passes ok, write will be performed to the database. If the validation phase fails to pass, all temporary written data will be aborted.
- 14. The Algorithm (Kung-Robinson) •Validation Phase (can be described as) –T: a transaction –ts: the highest sequence number at the start of T –tf: the highest sequence number at the beginning of its validation phase 14 Shounak Katyayan Advances in Operating System Design M. Tech CSE : 2014 - 15 valid:=true; for t:=ts+1 to tf do if (writeset(t) & readset[T] != {} ) then valid :=false; if valid then {write phase; increment counter, assign T a sequence number}
- 15. Concurrency Control Algorithms •Fully replicated database systems i.e. “data objects are replicated at all sites” 15 Shounak Katyayan Advances in Operating System Design M. Tech CSE : 2014 - 15
- 16. Completely Centralized Algorithm 16 Shounak Katyayan Advances in Operating System Design M. Tech CSE : 2014 - 15 Central Site Site A Site B Site C Site D Site E A site is designated as the central site.
- 17. Completely Centralized Algorithm 17 Shounak Katyayan Advances in Operating System Design M. Tech CSE : 2014 - 15 Central Site Site A Site B Site C Site D Site E Transaction: Update Account Balance #B=78000 Transaction at site A occurs.
- 18. Completely Centralized Algorithm 18 Shounak Katyayan Advances in Operating System Design M. Tech CSE : 2014 - 15 Central Site Site A Site B Site C Site D Site E Transaction: Update Account Balance #B=78000 Transaction is forward to the central site for execution.
- 19. Completely Centralized Algorithm 19 Shounak Katyayan Advances in Operating System Design M. Tech CSE : 2014 - 15 T1 T2 T3 T4 T5 Central Site Site A Site B Site C Site D Site E Serial execution effect.
- 20. Completely Centralized Algorithm 20 Shounak Katyayan Advances in Operating System Design M. Tech CSE : 2014 - 15 T1 T2 T3 T4 T5 Central Site Site A Site B Site C Site D Site E Serial execution effect.
- 21. Completely Centralized Algorithm 21 Shounak Katyayan Advances in Operating System Design M. Tech CSE : 2014 - 15 T1 T2 T3 T4 T5 Central Site Site A Site B Site C Site D Site E Serial execution effect.
- 22. Completely Centralized Algorithm 22 Shounak Katyayan Advances in Operating System Design M. Tech CSE : 2014 - 15 T1 T2 T3 T4 T5 Central Site Site A Site B Site C Site D Site E Serial execution effect.
- 23. Completely Centralized Algorithm 23 Shounak Katyayan Advances in Operating System Design M. Tech CSE : 2014 - 15 T1 T2 T3 T4 T5 Central Site Site A Site B Site C Site D Site E Serial execution effect.
- 24. Completely Centralized Algorithm 24 Shounak Katyayan Advances in Operating System Design M. Tech CSE : 2014 - 15 T1 T2 T3 T4 T5 Central Site Site A Site B Site C Site D Site E Perform update message broadcast to all other sites with sequence number. Perform Update Perform Update Perform Update Perform Update Perform Update
- 25. Completely Centralized Algorithm 25 Shounak Katyayan Advances in Operating System Design M. Tech CSE : 2014 - 15 T1 T2 T3 T4 T5 Central Site Site A Site B Site C Site D Site E Messages processed then updates applied to local databases. Perform Update Perform Update Perform Update Perform Update Perform Update
- 26. Centralized Locking Algorithm •Transactions processed in distributed manner •Central site is requested with a lock request message for data objects to be accessed •Site responds with a lock grant message with a sequence number •Queue for each data object •Site executes its transaction after receiving lock grant and broadcasts perform update after which the central site releases all locks 26 Shounak Katyayan Advances in Operating System Design M. Tech CSE : 2014 - 15
- 27. Centralized Locking Algorithm 27 Shounak Katyayan Advances in Operating System Design M. Tech CSE : 2014 - 15 Many Sites with centralized lock management Central Site
- 28. Centralized Locking Algorithm 28 Shounak Katyayan Advances in Operating System Design M. Tech CSE : 2014 - 15 Transactions processed at their home site (distributed manner).
- 29. Centralized Locking Algorithm 29 Shounak Katyayan Advances in Operating System Design M. Tech CSE : 2014 - 15 Site requests locks via lock request message for the data objects when it needs to update database lock request message O1 O2 O7 (objects)
- 30. Centralized Locking Algorithm 30 Shounak Katyayan Advances in Operating System Design M. Tech CSE : 2014 - 15 Central site responds with a lock grant message (if all locks can be granted) lock grant message O1 O2 O7 (objects) O3 O4 O5 O6 O8 O9
- 31. Centralized Locking Algorithm 31 Shounak Katyayan Advances in Operating System Design M. Tech CSE : 2014 - 15 Site executes its transaction after receiving lock grant O1 O2 O7 (objects) O3 O4 O5 O6 O8 O9
- 32. Centralized Locking Algorithm 32 Shounak Katyayan Advances in Operating System Design M. Tech CSE : 2014 - 15 Site broadcasts perform update message O1 O2 O7 (objects) Perform Update Perform Update Perform Update Perform Update Perform Update O3 O4 O5 O6 O8 O9
- 33. Centralized Locking Algorithm 33 Shounak Katyayan Advances in Operating System Design M. Tech CSE : 2014 - 15 Central site releases all locks when it receives perform update message O1 O2 O7 (objects) Perform Update Perform Update Perform Update Perform Update Perform Update O3 O4 O5 O6 O8 O9
- 34. INGRES’ Primary-Site Locking Algorithm •Based on primary site method •Lock management distributed among all the sites •Each object of database designated with a single primary site 34 Shounak Katyayan Advances in Operating System Design M. Tech CSE : 2014 - 15
- 35. Two Phase Locking (2PL) Algorithm •Two-phase locking protocol – Each transaction is executed in two phases ∗ Growing phase: the transaction obtains locks ∗ Shrinking phase: the transaction releases locks 35 Shounak Katyayan Advances in Operating System Design M. Tech CSE : 2014 - 15 The lock point is the moment when transitioning from the growing phase to the shrinking phase
- 36. Two Phase Locking (2PL) Algorithm •Properties of the 2PL protocol –Generates conflict-serializable schedules –But schedules may cause cascading aborts •If a transaction aborts after it releases a lock, it may cause other transactions that have accessed the unlocked data item to abort as well • Strict 2PL locking protocol Holds the locks till the end of the transaction Cascading aborts are avoided 36 Shounak Katyayan Advances in Operating System Design M. Tech CSE : 2014 - 15
- 37. Optimistic Algorithms & Concurrency Control Algorithms 37 Shounak Katyayan Advances in Operating System Design Shounak Katyayan M. Tech CSE : 2014 - 15 Thank You !!!