數(shù)據(jù)庫系統(tǒng)原理英文課件：ch16 Concurrency Control

1、Database System Concepts 5th Ed. Silberschatz, Korth and Sudarshan, 2005 See www.db- for conditions on re-use Silberschatz, Korth and Sudarshan16.2Database System Concepts - 5th Edition, Oct 5, 2006nLock-Based ProtocolsnTimestamp-Based ProtocolsnValidation-Based ProtocolsnMultiple GranularitynMultiv

2、ersion SchemesnInsert and Delete OperationsnConcurrency in Index StructuresSilberschatz, Korth and Sudarshan16.3Database System Concepts - 5th Edition, Oct 5, 2006nA lock is a mechanism to control concurrent access to a data itemnData items can be locked in two modes : 1. exclusive (X) mode. Data it

3、em can be both read as well as written. X-lock is requested using lock-X instruction. 2. shared (S) mode. Data item can only be read. S-lock is requested using lock-S instruction.nLock requests are made to concurrency-control manager. Transaction can proceed only after request is granted.Silberschat

4、z, Korth and Sudarshan16.4Database System Concepts - 5th Edition, Oct 5, 2006nLock-compatibility matrixnA transaction may be granted a lock on an item if the requested lock is compatible with locks already held on the item by other transactionsnAny number of transactions can hold shared locks on an

5、item, lbut if any transaction holds an exclusive on the item no other transaction may hold any lock on the item.nIf a lock cannot be granted, the requesting transaction is made to wait till all incompatible locks held by other transactions have been released. The lock is then granted.Silberschatz, K

6、orth and Sudarshan16.5Database System Concepts - 5th Edition, Oct 5, 2006nExample of a transaction performing locking: T2: lock-S(A); read (A); unlock(A); lock-S(B); read (B); unlock(B); display(A+B)nLocking as above is not sufficient to guarantee serializability if A and B get updated in-between th

7、e read of A and B, the displayed sum would be wrong.nA locking protocol is a set of rules followed by all transactions while requesting and releasing locks. Locking protocols restrict the set of possible schedules.Silberschatz, Korth and Sudarshan16.6Database System Concepts - 5th Edition, Oct 5, 20

8、06nConsider the partial schedulenNeither T3 nor T4 can make progress executing lock-S(B) causes T4 to wait for T3 to release its lock on B, while executing lock-X(A) causes T3 to wait for T4 to release its lock on A.nSuch a situation is called a deadlock. lTo handle a deadlock one of T3 or T4 must b

9、e rolled back and its locks released.Silberschatz, Korth and Sudarshan16.7Database System Concepts - 5th Edition, Oct 5, 2006nThe potential for deadlock exists in most locking protocols. Deadlocks are a necessary evil.nStarvation is also possible if concurrency control manager is badly designed. For

10、 example:lA transaction may be waiting for an X-lock on an item, while a sequence of other transactions request and are granted an S-lock on the same item. lThe same transaction is repeatedly rolled back due to deadlocks.nConcurrency control manager can be designed to prevent starvation.Silberschatz

11、, Korth and Sudarshan16.8Database System Concepts - 5th Edition, Oct 5, 2006nThis is a protocol which ensures conflict-serializable schedules.nPhase 1: Growing Phaseltransaction may obtain locks ltransaction may not release locksnPhase 2: Shrinking Phaseltransaction may release locksltransaction may

12、 not obtain locksnThe protocol assures serializability. It can be proved that the transactions can be serialized in the order of their lock points (i.e. the point where a transaction acquired its final lock). Silberschatz, Korth and Sudarshan16.9Database System Concepts - 5th Edition, Oct 5, 2006nTw

13、o-phase locking does not ensure freedom from deadlocksnCascading roll-back is possible under two-phase locking. To avoid this, follow a modified protocol called strict two-phase locking. Here a transaction must hold all its exclusive locks till it commits/aborts.nRigorous two-phase locking is even s

14、tricter: here all locks are held till commit/abort. In this protocol transactions can be serialized in the order in which they commit.Silberschatz, Korth and Sudarshan16.10Database System Concepts - 5th Edition, Oct 5, 2006nThere can be conflict serializable schedules that cannot be obtained if two-

15、phase locking is used. nHowever, in the absence of extra information (e.g., ordering of access to data), two-phase locking is needed for conflict serializability in the following sense: Given a transaction Ti that does not follow two-phase locking, we can find a transaction Tj that uses two-phase lo

16、cking, and a schedule for Ti and Tj that is not conflict serializable.Silberschatz, Korth and Sudarshan16.11Database System Concepts - 5th Edition, Oct 5, 2006nTwo-phase locking with lock conversions: First Phase: lcan acquire a lock-S on itemlcan acquire a lock-X on itemlcan convert a lock-S to a l

17、ock-X (upgrade) Second Phase:lcan release a lock-Slcan release a lock-Xlcan convert a lock-X to a lock-S (downgrade)nThis protocol assures serializability. But still relies on the programmer to insert the various locking instructions.Silberschatz, Korth and Sudarshan16.12Database System Concepts - 5

18、th Edition, Oct 5, 2006nA transaction Ti issues the standard read/write instruction, without explicit locking calls.nThe operation read(D) is processed as: if Ti has a lock on D then read(D) else begin if necessary wait until no other transaction has a lock-X on D grant Ti a lock-S on D; read(D) end

19、Silberschatz, Korth and Sudarshan16.13Database System Concepts - 5th Edition, Oct 5, 2006nwrite(D) is processed as: if Ti has a lock-X on D then write(D) else begin if necessary wait until no other trans. has any lock on D, if Ti has a lock-S on D then upgrade lock on D to lock-X else grant Ti a loc

20、k-X on D write(D) end;nAll locks are released after commit or abortSilberschatz, Korth and Sudarshan16.14Database System Concepts - 5th Edition, Oct 5, 2006nA lock manager can be implemented as a separate process to which transactions send lock and unlock requestsnThe lock manager replies to a lock

21、request by sending a lock grant messages (or a message asking the transaction to roll back, in case of a deadlock)nThe requesting transaction waits until its request is answerednThe lock manager maintains a data-structure called a lock table to record granted locks and pending requestsnThe lock tabl

22、e is usually implemented as an in-memory hash table indexed on the name of the data item being lockedSilberschatz, Korth and Sudarshan16.15Database System Concepts - 5th Edition, Oct 5, 2006nBlack rectangles indicate granted locks, white ones indicate waiting requestsnLock table also records the typ

23、e of lock granted or requestednNew request is added to the end of the queue of requests for the data item, and granted if it is compatible with all earlier locksnUnlock requests result in the request being deleted, and later requests are checked to see if they can now be grantednIf transaction abort

24、s, all waiting or granted requests of the transaction are deleted llock manager may keep a list of locks held by each transaction, to implement this efficientlyGrantedWaitingSilberschatz, Korth and Sudarshan16.16Database System Concepts - 5th Edition, Oct 5, 2006nGraph-based protocols are an alterna

25、tive to two-phase lockingnImpose a partial ordering on the set D = d1, d2 ,., dh of all data items.lIf di dj then any transaction accessing both di and dj must access di before accessing dj.lImplies that the set D may now be viewed as a directed acyclic graph, called a database graph.nThe tree-proto

26、col is a simple kind of graph protocol. Silberschatz, Korth and Sudarshan16.17Database System Concepts - 5th Edition, Oct 5, 20061.Only exclusive locks are allowed.2.The first lock by Ti may be on any data item. Subsequently, a data Q can be locked by Ti only if the parent of Q is currently locked b

27、y Ti.3.Data items may be unlocked at any time.4.A data item that has been locked and unlocked by Ti cannot subsequently be relocked by Ti Silberschatz, Korth and Sudarshan16.18Database System Concepts - 5th Edition, Oct 5, 2006nThe tree protocol ensures conflict serializability as well as freedom fr

28、om deadlock.nUnlocking may occur earlier in the tree-locking protocol than in the two-phase locking protocol.lshorter waiting times, and increase in concurrencylprotocol is deadlock-free, no rollbacks are requirednDrawbackslProtocol does not guarantee recoverability or cascade freedom4Need to introd

29、uce commit dependencies to ensure recoverability lTransactions may have to lock data items that they do not access.4increased locking overhead, and additional waiting time4potential decrease in concurrencynSchedules not possible under two-phase locking are possible under tree protocol, and vice vers

30、a.Silberschatz, Korth and Sudarshan16.19Database System Concepts - 5th Edition, Oct 5, 2006nAllow data items to be of various sizes and define a hierarchy of data granularities, where the small granularities are nested within larger onesnCan be represented graphically as a tree (but dont confuse wit

31、h tree-locking protocol)nWhen a transaction locks a node in the tree explicitly, it implicitly locks all the nodes descendents in the same mode.nGranularity of locking (level in tree where locking is done):lfine granularity (lower in tree): high concurrency, high locking overheadlcoarse granularity

32、(higher in tree): low locking overhead, low concurrencySilberschatz, Korth and Sudarshan16.20Database System Concepts - 5th Edition, Oct 5, 2006 The levels, starting from the coarsest (top) level areldatabaselarealfilelrecord Silberschatz, Korth and Sudarshan16.21Database System Concepts - 5th Editi

33、on, Oct 5, 2006nIn addition to S and X lock modes, there are three additional lock modes with multiple granularity:lintention-shared (IS): indicates explicit locking at a lower level of the tree but only with shared locks.lintention-exclusive (IX): indicates explicit locking at a lower level with ex

34、clusive or shared lockslshared and intention-exclusive (SIX): the subtree rooted by that node is locked explicitly in shared mode and explicit locking is being done at a lower level with exclusive-mode locks.nintention locks allow a higher level node to be locked in S or X mode without having to che

35、ck all descendent nodes.Silberschatz, Korth and Sudarshan16.22Database System Concepts - 5th Edition, Oct 5, 2006nThe compatibility matrix for all lock modes is: ISIXSS IXX ISIXSS IXX Silberschatz, Korth and Sudarshan16.23Database System Concepts - 5th Edition, Oct 5, 2006nTransaction Ti can lock a

36、node Q, using the following rules:1.The lock compatibility matrix must be observed.2.The root of the tree must be locked first, and may be locked in any mode.3.A node Q can be locked by Ti in S or IS mode only if the parent of Q is currently locked by Ti in either IX or IS mode.4.A node Q can be loc

37、ked by Ti in X, SIX, or IX mode only if the parent of Q is currently locked by Ti in either IX or SIX mode.5.Ti can lock a node only if it has not previously unlocked any node (that is, Ti is two-phase).6.Ti can unlock a node Q only if none of the children of Q are currently locked by Ti.nObserve th

38、at locks are acquired in root-to-leaf order, whereas they are released in leaf-to-root order.Silberschatz, Korth and Sudarshan16.24Database System Concepts - 5th Edition, Oct 5, 2006nConsider the following two transactions: T1: write (X) T2: write(Y) write(Y) write(X)nSchedule with deadlockT1T2lock-

39、X on Xwrite (X) lock-X on Ywrite (X) wait for lock-X on Xwait for lock-X on YSilberschatz, Korth and Sudarshan16.25Database System Concepts - 5th Edition, Oct 5, 2006nSystem is deadlocked if there is a set of transactions such that every transaction in the set is waiting for another transaction in t

40、he set.nDeadlock prevention protocols ensure that the system will never enter into a deadlock state. Some prevention strategies :lRequire that each transaction locks all its data items before it begins execution (predeclaration).lImpose partial ordering of all data items and require that a transacti

41、on can lock data items only in the order specified by the partial order (graph-based protocol).Silberschatz, Korth and Sudarshan16.26Database System Concepts - 5th Edition, Oct 5, 2006nFollowing schemes use transaction timestamps for the sake of deadlock prevention alone.nwait-die scheme non-preempt

42、ivelolder transaction may wait for younger one to release data item. Younger transactions never wait for older ones; they are rolled back instead.la transaction may die several times before acquiring needed data itemnwound-wait scheme preemptivelolder transaction wounds (forces rollback) of younger

43、transaction instead of waiting for it. Younger transactions may wait for older ones.lmay be fewer rollbacks than wait-die scheme.Silberschatz, Korth and Sudarshan16.27Database System Concepts - 5th Edition, Oct 5, 2006nBoth in wait-die and in wound-wait schemes, a rolled back transactions is restart

44、ed with its original timestamp. Older transactions thus have precedence over newer ones, and starvation is hence avoided.nTimeout-Based Schemes :la transaction waits for a lock only for a specified amount of time. After that, the wait times out and the transaction is rolled back.lthus deadlocks are

45、not possiblelsimple to implement; but starvation is possible. Also difficult to determine good value of the timeout interval.Silberschatz, Korth and Sudarshan16.28Database System Concepts - 5th Edition, Oct 5, 2006nDeadlocks can be described as a wait-for graph, which consists of a pair G = (V,E), l

46、V is a set of vertices (all the transactions in the system)lE is a set of edges; each element is an ordered pair Ti Tj. nIf Ti Tj is in E, then there is a directed edge from Ti to Tj, implying that Ti is waiting for Tj to release a data item.nWhen Ti requests a data item currently being held by Tj,

47、then the edge Ti Tj is inserted in the wait-for graph. This edge is removed only when Tj is no longer holding a data item needed by Ti.nThe system is in a deadlock state if and only if the wait-for graph has a cycle. Must invoke a deadlock-detection algorithm periodically to look for cycles.Silbersc

48、hatz, Korth and Sudarshan16.29Database System Concepts - 5th Edition, Oct 5, 2006Wait-for graph without a cycleWait-for graph with a cycleSilberschatz, Korth and Sudarshan16.30Database System Concepts - 5th Edition, Oct 5, 2006nWhen deadlock is detected :lSome transaction will have to rolled back (m

49、ade a victim) to break deadlock. Select that transaction as victim that will incur minimum cost.lRollback - determine how far to roll back transaction4Total rollback: Abort the transaction and then restart it.4More effective to roll back transaction only as far as necessary to break deadlock.lStarva

50、tion happens if same transaction is always chosen as victim. Include the number of rollbacks in the cost factor to avoid starvationDatabase System Concepts 5th Ed. Silberschatz, Korth and Sudarshan, 2005 See www.db- for conditions on re-use Silberschatz, Korth and Sudarshan16.32Database System Conce

51、pts - 5th Edition, Oct 5, 2006nEach transaction is issued a timestamp when it enters the system. If an old transaction Ti has time-stamp TS(Ti), a new transaction Tj is assigned time-stamp TS(Tj) such that TS(Ti) TS(Tj). nThe protocol manages concurrent execution such that the time-stamps determine

52、the serializability order.nIn order to assure such behavior, the protocol maintains for each data Q two timestamp values:lW-timestamp(Q) is the largest time-stamp of any transaction that executed write(Q) successfully.lR-timestamp(Q) is the largest time-stamp of any transaction that executed read(Q)

53、 successfully.Silberschatz, Korth and Sudarshan16.33Database System Concepts - 5th Edition, Oct 5, 2006nThe timestamp ordering protocol ensures that any conflicting read and write operations are executed in timestamp order.nSuppose a transaction Ti issues a read(Q)1.If TS(Ti) W-timestamp(Q), then Ti

54、 needs to read a value of Q that was already overwritten.nHence, the read operation is rejected, and Ti is rolled back.2.If TS(Ti) W-timestamp(Q), then the read operation is executed, and R-timestamp(Q) is set to max(R-timestamp(Q), TS(Ti).Silberschatz, Korth and Sudarshan16.34Database System Concep

55、ts - 5th Edition, Oct 5, 2006nSuppose that transaction Ti issues write(Q).1.If TS(Ti) R-timestamp(Q), then the value of Q that Ti is producing was needed previously, and the system assumed that that value would never be produced. nHence, the write operation is rejected, and Ti is rolled back.2.If TS

56、(Ti) W-timestamp(Q), then Ti is attempting to write an obsolete value of Q. nHence, this write operation is rejected, and Ti is rolled back.3.Otherwise, the write operation is executed, and W-timestamp(Q) is set to TS(Ti).Silberschatz, Korth and Sudarshan16.35Database System Concepts - 5th Edition,

57、Oct 5, 2006A partial schedule for several data items for transactions withtimestamps 1, 2, 3, 4, 5T1T2T3T4T5read(Y)read(X) read(Y)write(Y) write(Z) read(Z) read(X) abort read(X) write(Z) abort write(Y) write(Z) Silberschatz, Korth and Sudarshan16.36Database System Concepts - 5th Edition, Oct 5, 2006

58、nThe timestamp-ordering protocol guarantees serializability since all the arcs in the precedence graph are of the form: Thus, there will be no cycles in the precedence graphnTimestamp protocol ensures freedom from deadlock as no transaction ever waits. nBut the schedule may not be cascade-free, and

59、may not even be recoverable.transactionwith smallertimestamptransactionwith largertimestamp Silberschatz, Korth and Sudarshan16.37Database System Concepts - 5th Edition, Oct 5, 2006nProblem with timestamp-ordering protocol:lSuppose Ti aborts, but Tj has read a data item written by TilThen Tj must ab

60、ort; if Tj had been allowed to commit earlier, the schedule is not recoverable.lFurther, any transaction that has read a data item written by Tj must abortlThis can lead to cascading rollback - that is, a chain of rollbacks n Solution 1:lA transaction is structured such that its writes are all perfo