Concurrency Control

CS 728Advanced Database SystemsChapter 21 Introduction to Protocols for Concurrency Control in Databases

Concurrency Control • The main aim of any Database Management System is to control requests for the same data, at the same time, from multiple users. • Concurrency control algorithms try to coordinate the operations of concurrent transactions to prevent interference among concurrently executing transactions in order to achieve transaction consistency.

Concurrency Control • Purpose of Concurrency Control • To enforce Isolation (through mutual exclusion) among conflicting transactions. • To preserve database consistency through consistency preserving execution of transactions. • To resolve read-write and write-write conflicts. • Example: • In concurrent execution environment if T1 conflicts with T2 over a data item A, then the existing concurrency control decides if T1 or T2 should get the A and if the other transaction is rolled-back or waits.

Concurrency control ensures that transactions are updated in the correct order, i.e. it ensures the serializability of transactions in a multi-user database environment. One way to insure serializability is to allow a transaction to access a data item only if it is currently holding a lock on that item. A lock is a variable associated with a data item that describes the status of the item with respect to possible operations that can be applied to it. Generally, there is one lock for each data item in the database. Lock requests are made to concurrency-control manager Transaction can proceed only after request is granted Lock-Based Protocols

Lock-Based Protocols • Locking is an operation which secures permission to Read and Write a data item for a transaction. • Example: • Lock (X): Data item X is locked in behalf of the requesting transaction. • Unlocking is an operation which removes these permissions from the data item. • Example: • Unlock (X): Data item X is made available to all other transactions. • Lock and Unlock are Atomic operations.

Shared/Exclusive Locks • Data items can be locked in two modes : • Shared (read) mode: read_lock(Q)/lock-S(Q) • More than one transaction can apply share lock on Q for reading its value but no write lock can be applied on Q by any other transaction. • Exclusive (write) mode: write_lock(Q)/lock-X(Q) • Only one write lock on Q can exist at any time and no shared lock can be applied by any other transaction on Q. • Conflict matrix • Lock-compatibility matrix

A transaction may be granted a lock on an item if the requested lock is compatible with locks already held on the item by other transactions Any number of transactions can hold shared locks on an item, but if any transaction holds an exclusive on the item no other transaction may hold any lock on the item If 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. Shared/Exclusive Locks

Shared/Exclusive Locks • In shared/exclusive locking system, every transaction must obey the following rules: 1. Issue the operation lock-S(Q) or lock-X(Q) before any read(Q) operation, 2. Issue the operation lock-X(Q) before any write(Q) operation is performed, 3. Issue the operation unlock(Q) after all read(Q) and write(Q) operations, 4. Not issue a lock-S(Q) operation if it already holds an exclusive lock on item Q, 5. Not issue a lock-X(Q) operation if it already holds a shared lock or exclusive lock on item Q, 6. Not issue an unlock(Q) operation unless it already holds a read lock or write lock on item Q.

Lock-Based Protocols • The following code performs the read_lock(X) operation: B: if LOCK(X) = ”unlocked” then begin LOCK(X) = “read-locked”; no_of_reads(X) = 1; end else if LOCK(X) = “read-locked” then no_of_reads(X)++ else begin wait(until LOCK(X)=”unlocked” and the lock manager wakes up the transaction); go to B end;

Lock-Based Protocols • The following code performs the write_lock(X) operation: B: if LOCK(X) = “unlocked” then LOCK(X) = “write-locked”; else begin wait (until LOCK(X)=“unlocked” and the lock manager wakes up the transaction); go to B end;

Lock-Based Protocols • The following code performs the unlock operation: if LOCK (X) = “write-locked” then begin LOCK (X)  “unlocked”; wakes up one of the transactions, if any end else if LOCK (X)  “read-locked” then begin no_of_reads(X)  no_of_reads(X)-1 if no_of_reads (X) = 0 then begin LOCK (X) = “unlocked”; wake up one of the transactions, if any end end;

Consider the partial schedule Neither 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. Pitfalls of Lock-Based Protocols

Such a situation is called a deadlock. To handle a deadlock, one of T3 or T4 must be rolled back and its locks released. The potential for deadlock exists in most locking protocols. Deadlocks are a necessary evil. Pitfalls of Lock-Based Protocols

Dealing with Deadlock • A partial schedule of T1 and T2  that is in a state of deadlock. • A wait-for graph for the partial schedule in (a).

Dealing with Starvation • Starvation is the situation in which a transaction cannot proceed for an indefinite period of time while other transactions in the system continue normally. • For example: • A 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. • The same transaction is repeatedly rolled back due to deadlocks. • One solution for starvation is to have a fair scheme, such as using a first-come-first-served.

This protocol ensures conflict-serializable schedules If all transactions obey the 2PL then all possible interleaved schedules are serializable. This protocol requires that each transaction issues lock and unlock requests in two phases: Phase 1: Growing Phase transaction may obtain locks but may not release locks Phase 2: Shrinking Phase transaction may release locks but may not obtain locks The Two-Phase Locking Protocol

No transaction should request a lock after it releases one of its locks. The Two-Phase Locking Protocol

It can be proved that the transactions can be serialized in the order of their lock points a point where a transaction acquired its final lock end of its growing phase 2PL does not ensure freedom from deadlocks The Two-Phase Locking Protocol

Two-Phase Locking • Transactions that do not obey 2PL • Two transactions T1 and T2. • Results of possible serial schedules of T1 and T2. T2 T1

Two-Phase Locking • Transactions that do not obey 2PL • (c) A nonserializable schedule S that uses locks.

Two-Phase Locking • Transactions T1 & T2, which are the same as T1 & T2 but which follow 2PL.

To avoid cascading rollback, follow a modified protocol called Strict 2PL (S2PL). 2PL a transaction must hold all its exclusive locks till it commits/aborts. S2PL does not prevent deadlock. Strict Two-Phase Locking (S2PL) locks time T commits

R2PL is even stricter all locks are held till commit/abort. In this protocol, transactions can be serialized in the order in which they commit. S2PL permits higher degree of concurrency than R2PL but less than 2PL. Rigorous two-phase locking (R2PL)

2PL with lock conversions: First Phase (Growing): can acquire a lock-S/lock-X on item can convert a lock-S to a lock-X (upgrade) if Ti has a read-lock (X) and Tj has no read-lock (X) (i  j) then convert read-lock(X) to write-lock(X) Else force Ti to wait until Tj unlocks X Second Phase (Shrinking): can release a lock-S/lock-X can convert a lock-X to a lock-S (downgrade) Ti has a write-lock(X) (*no transaction can have any lock on X*) convert write-lock(X) to read-lock(X) This protocol assures serializability Lock Conversions

Each transaction is issued a timestamp when it starts CC techniques based on timestamp ordering do no use locks, and thus deadlocks cannot occur (no transaction ever waits) may not be (cascadeless and recoverable) The protocol manages concurrent execution such that the time-stamps determine the serializability order. 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) Timestamp-Based Protocols

In order to assure such behavior, the protocol maintains for each data Q two timestamp values: W-timestamp(Q) (W-TS(Q)) is the largest time-stamp of any transaction that executed write(Q) successfully. If W-TS(Q) = TS(T), then T is the youngest transaction that has written Q successfully. R-timestamp(Q) (R-TS(Q)) is the largest time-stamp of any transaction that executed read(Q) successfully. If R-TS(Q) = TS(T), then T is the youngest transaction that has read Q successfully. These TSs are updated whenever a new read(Q) or write(Q) is executed Timestamp-Ordering Protocol

The timestamp ordering protocol ensures that any conflicting read and write operations are executed in timestamp order Suppose a transaction T issues read(Q) If TS(T) W-TS(Q) then T needs to read a value of Q that was already overwritten. Hence, the read operation is rejected, and T is rolled back. If TS(T) W-TS(Q), then the read operation is executed, and R-TS(Q) = max(R-TS(Q), TS(T)) Timestamp-Ordering Protocol

Suppose that transaction T issues write(Q). If TS(T) R-TS(Q), then the value of Q that T is producing was needed previously, and the system assumed that the value would never be produced. Hence, the write operation is rejected, and T is rolled back. If TS(T) W-TS(Q), then T is attempting to write an obsolete value of Q. Hence, this write operation is rejected, and T is rolled back. Otherwise, the write operation is executed, and W-TS(Q) = TS(T) Timestamp-Ordering Protocol

Example Use of the Protocol R-TS(A) W-TS(A) T1 T2 T3 0 0 read(A) 1 0 read(A) 3 0 read(A) 3 0 write(A) rejected • TS(T1) = 1, TS(T2) = 2, TS(T3) = 3

Transactions timestamps are 1, 2, 3, 4, 5 Example Use of the Protocol T1 T2 T3 T4 T5 read(X) read(Y) read(Y) write(Y) write(Z) read(Z) read(X) read(X) write(Z) reject & roll back write(Y) write(Z)

Correctness of Timestamp-Ordering Protocol transaction with smaller timestamp transaction with larger timestamp • The 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 graph

Problem with timestamp-ordering protocol: Suppose Ti aborts, but Tj has read a data item written by Ti, then Tjmust abort If Tjhad been allowed to commit earlier, the schedule is not recoverable. Further, any transaction that has read a data item written by Tj must abort This can lead to cascading rollback a chain of rollbacks Recoverability and Cascade Freedom

Solution: A transaction is structured such that its writes are all performed at the end of its processing All writes of a transaction form an atomic action; no transaction may execute while a transaction is being written A transaction that aborts is restarted with a new timestamp Recoverability and Cascade Freedom

Modified version of the timestamp-ordering protocol in which obsolete (outdated) write (the value that will never need to be read) operations may be ignored under certain circumstances. When T attempts to write data item Q, if TS(T)  W-TS(Q), then T is attempting to write an obsolete value of Q. Hence, rather than rolling back T as the timestamp ordering protocol would have done, this write operation can be ignored. Otherwise this protocol is the same as the timestamp ordering protocol. Thomas' Write Rule allows greater potential concurrency. Thomas’ Write Rule

Consider the following two transactions: T1: write (A) T2: write(B) write(B) write(A) Schedule with deadlock Deadlock Handling

Deadlock Prevention • Deadlock prevention protocols ensure that the system will never enter into a deadlock state. • Some prevention strategies : • Requires that each transaction locks all its data items before it begins execution. • Low degree of concurrency

Deadlock Prevention • Conservative 2 PL (static & deadlock-free): • requires a transaction T to pre-declare all the read & write set of items; and lock all these items before T begins execution. • If any of the pre-declared items can not be locked, T does not lock any item at all. Instead, T waits and tries again until all the items are available for locking.

Deadlock Prevention • Assume that Ti requests a data item currently held by Tj. • wait-die scheme: • If Ti is older than Tj (i.e., TS(Ti)  TS(Tj)) • Then wait(Ti) • Else die(Ti) • Ti is aborted and restarted with its old starting time. • Younger transactions never wait for older ones; they are rolled back instead. • A transaction may die several times before acquiring needed data item

wait? Wait-Die: Example T1 (ts =10) T2 (ts =20) T3 (ts =25) wait wait

Deadlock Prevention • wound-wait scheme • If Ti is older than Tj (i.e., TS(Ti)  TS(Tj)) • then wound(Tj) // Tj is wounded by Ti • Tj is aborted and restart it with its old starting time. • else (Ti is younger than Tj) wait(Ti) • Older transaction wounds (forces rollback) of younger transaction instead of waiting for it. • Younger transactions may wait for older ones. • May be fewer rollbacks than wait-die scheme.

Older transactions thus have precedence over newer ones, and starvation is hence avoided. Example: T1: W(X) W(Y) T2: W(Y) W(X) T1 is older. wait-die: X-Lock1(X) X-Lock2(Y) wait(T1,Y) … wound-wait: X-Lock1(X) X-Lock2(Y) abort(T2) X-Lock1(Y) … Deadlock Prevention

wait Wound-Wait: Example T1 (ts =25) T2 (ts =20) T3 (ts =10) wait wait

If a transaction waits for a lock more than a specified amount of time, the transaction is rolled back. deadlocks are not possible simple to implement starvation is possible difficult to select a good timeout value Timeout-Based Schemes

Deadlocks can be described as a wait-for graph, which consists of a pair G = (V, E), V is a set of vertices (all the transactions) E is a set of edges each element is an ordered pair TiTj. If TiTjis in E, then Ti is waiting for Tj to release a data item. When Ti requests a data item currently being held by Tj, then the edge TiTjis inserted in the wait-for graph. This edge is removed only when Tj is no longer holding a data item needed by Ti. Deadlock Detection

The 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. Deadlock Detection Wait-for graph without a cycle Wait-for graph with a cycle

When deadlock is detected : Some transaction will have to rolled back (made a victim) to break deadlock. Select that transaction as victim that will incur minimum cost. Factors in selecting a victim transaction: The amount of effort already made in the transaction. The cost of aborting the transaction. It may cause cascading aborts. How close the transaction is to complete? The number of deadlocks that can be broken when the transaction is aborted. Deadlock Recovery

When deadlock is detected: Rollback: determine how far to rollback transaction Total rollback: Abort the transaction and then restart it. Partial rollback: More effective to roll back transaction only as far as necessary to break deadlock. Starvation happens if same transaction is always chosen as victim. Include the number of rollbacks in the cost factor to avoid starvation Deadlock Recovery

Concurrency Control