DBMS

UNIT 4
 UNIT-4
Transactions, Concurrency
Control and Recovery
Transaction Management
Transaction Management
 Transaction
 Transaction Concept
 Transaction State
 Concurrent Executions
 Serializability
 Recoverability
 Testing for Serializability
 Concurrency control
 Lock-based protocols
 Timestamp-based protocols
 Multiple granularity
 Multiversion schemes
 Recovery Systems
 Log-based recovery
 Recovery with concurrent transactions
Transactions:
 A transaction is a collection of actions that make consistent transformations of system
states.
 Formally, it is a partial order over the operations that are part of the transactions.
Examples-
 Consider an airline reservation example:
 Transaction Reservation
begin
input (flight, date, customer_name);
temp <- Read(flight(date).sold_seats);
Write(flight(date).sold_seats, temp + 1);
Write(flight(date).cust_name, customer_name);
Output ("reservation completed")
end. {transaction}
What if there are no free seats?
 Transaction Reservation
begin
input (flight, date, customer_name);
temp <- Read(flight(date).sold_seats);
if temp = flight(date).maximum then
begin
output("no free seats");
Abort

end
else begin
Write(flight(date).sold_seats, temp + 1);
Write(flight(date).cust_name, customer_name);
Commit;
output("reservation completed")

end
end. {transaction}
Transaction Concept
 A transaction is a unit of program execution that accesses and possibly updates
various data items.
 E.g. transaction to transfer €50 from account A to account B:
1. read_from_acoount(A)
2. A := A – 50
3. write_to_account(A)
4. read_from_accont(B)
5. B := B + 50
6. write_to_account(B)
 Two main issues to deal with:
 Failures of various kinds, such as hardware failures and system crashes
 Concurrent execution of multiple transactions
Transaction ACID properties
 E.g. transaction to transfer €50 from account A to account B:
1. read_from_acoount(A)
2. A := A – 50
3. write_to_account(A)
4. read_from_accont(B)
5. B := B + 50
6. write_to_account(B)
 Atomicity requirement
 if the transaction fails after step 3 and before step 6, money will be “lost” leading to an
inconsistent database state
 Failure could be due to software or hardware
 the system should ensure that updates of a partially executed transaction are not
reflected in the database
 All or nothing, regarding the execution of the transaction
 Durability requirement — once the user has been notified of transaction has completion, the
updates must persist in the database even if there are software or hardware failures.
Transaction ACID properties (Cont.)
 Transaction to transfer €50 from account A to account B:
1. read_from_acoount(A)
2. A := A – 50
3. write_to_account(A)
4. read_from_accont(B)
5. B := B + 50
6. write_to_account(B)
 Consistency requirement in above example:
 the sum of A and B is unchanged by the execution of the transaction
 In general, consistency requirements include
 Explicitly specified integrity constraints such as primary keys and foreign keys
 Implicit integrity constraints
 e.g. sum of balances of all accounts, minus sum of loan amounts must equal value of
cash-in-hand
 A transaction must see a consistent database and must leave a consistent database
 During transaction execution the database may be temporarily inconsistent.
 Constraints to be verified only at the end of the transaction
Transaction ACID properties (Cont.)
 Isolation requirement — if between steps 3 and 6, another transaction T2 is
allowed to access the partially updated database, it will see an inconsistent
database (the sum A + B will be less than it should be).
T1 T2
1. read(A)
2. A := A – 50
3. write(A)
read(A), read(B), print(A+B)
4. read(B)
5. B := B + 50
6. write(B
 Isolation can be ensured trivially by running transactions serially
 that is, one after the other.
 However, executing multiple transactions concurrently has significant benefits, as
we will see later.
ACID Properties - Summary
A transaction is a unit of program execution that accesses and
possibly updates various data items.To preserve the integrity of
data the database system must ensure:

 Atomicity Either all operations of the transaction are properly reflected in the
database or none are.
 Consistency Execution of a (single) transaction preserves the consistency of
the database.
 Isolation Although multiple transactions may execute concurrently, each
transaction must be unaware of other concurrently executing transactions.
Intermediate transaction results must be hidden from other concurrently
executed transactions.
 That is, for every pair of transactions Ti and Tj, it appears to Ti that either
Tj, finished execution before Ti started, or Tj started execution after Ti
finished.
 Durability. After a transaction completes successfully, the changes it has
made to the database persist, even if there are system failures.
Non-ACID Transactions
 There are application domains where ACID properties are not necessarily desired
or, most likely, not always possible.
 This is the case of so-called long-duration transactions
 Suppose that a transaction takes a lot of time
 In this case it is unlikely that isolation can/should be guaranteed
 E.g. Consider a transaction of booking a hotel and a flight
 Without Isolation, Atomicity may be compromised
 Consistency and Durability should be preserved
 Usual solution for long-duration transaction is to define compensation action –
what to do if later the transaction fails
 In (centralized) databases long-duration transactions are usually not considered.
 But these are more and more important, specially in the context of the Web.
Transaction State
 Active – the initial state; the transaction stays in this state while it is executing
 Partially committed – after the final statement has been executed.
 Failed – after the discovery that normal execution can no longer proceed.
 Aborted – after the transaction has been rolled back and the database restored to
its state prior to the start of the transaction. Two options after it has been aborted:
 restart the transaction
 can be done only if no internal logical error
 kill the transaction
 Committed – after successful completion.

 To guarantee atomicity, external observable action should all be performed (in

order) after the transaction is committed.
Transaction State (Cont.)
Implementation of Atomicity and Durability
 The recovery-management component of a database system
implements the support for atomicity and durability.
 E.g. the shadow-database scheme:
 all updates are made on a shadow copy of the database
 db_pointer is made to point to the updated shadow copy after
 the transaction reaches partial commit and
 all updated pages have been flushed to disk.
 Implementation of Atomicity and Durability (Cont.)
 db_pointer always points to the current consistent copy of the database.
 In case transaction fails, old consistent copy pointed to by db_pointer can be
used, and the shadow copy can be deleted.
 The shadow-database scheme:
 Assumes that only one transaction is active at a time.
 Assumes disks do not fail
 Useful for text editors, but
 extremely inefficient for large databases(!)
 Variant called shadow paging reduces copying of data, but is still not
practical for large databases
 Does not handle concurrent transactions
 Other implementations of atomicity and durability are possible, e.g. by using logs.
 Log-based recovery will be addressed later.
Concurrent Executions
 Multiple transactions are allowed to run concurrently in the system. Advantages
are:
 increased processor and disk utilization, leading to better transaction
throughput
 E.g. one transaction can be using the CPU while another is reading from or
writing to the disk
 reduced average response time for transactions: short transactions need not
wait behind long ones.
 Concurrency control schemes – mechanisms to achieve isolation
 that is, to control the interaction among the concurrent transactions in order
to prevent them from destroying the consistency of the database
 Two-phase look protocol
 Timestamp-Based Protocols
 Validation-Based Protocols
 Studied in Operating Systems, and briefly summarized later
Schedules
 Schedule – a sequences of instructions that specify the chronological order in which
instructions of concurrent transactions are executed
 a schedule for a set of transactions must consist of all instructions of those
transactions
 must preserve the order in which the instructions appear in each individual
transaction.
 A transaction that successfully completes its execution will have a commit
instructions as the last statement
 by default transaction assumed to execute commit instruction as its last step
 A transaction that fails to successfully complete its execution will have an abort
instruction as the last statement

 The goal is to find schedules that preserve the consistency.

Example Schedule 1
 Let T1 transfer €50 from A to B, and T2 transfer 10% of the balance from A to B.
 A serial schedule in which T1 is followed by T2 :
Example Schedule 2
• A serial schedule where T2 is followed by T1
Example Schedule 3
 Let T1 and T2 be the transactions defined previously. The following schedule
is not a serial schedule, but it is equivalent to Schedule 1.

In Schedules 1, 2 and 3, the sum A + B is preserved.

Example Schedule 4
 The following concurrent schedule does not preserve the value of (A + B ).
Serializability
 Goal : Deal with concurrent schedules that are equivalent to some serial execution:
 Basic Assumption – Each transaction preserves database consistency.
 Thus serial execution of a set of transactions preserves database consistency.
 A (possibly concurrent) schedule is serializable if it is equivalent to a serial
schedule. Different forms of schedule equivalence give rise to the notions of:
1. conflict serializability
2. view serializability
 Simplified view of transactions
 We ignore operations other than read and write instructions
 We assume that transactions may perform arbitrary computations on data in
local buffers in between reads and writes.
 Our simplified schedules consist of only read and write instructions.
Conflicting Instructions
 Instructions li and lj of transactions Ti and Tj respectively, conflict if and only if there
exists some item Q accessed by both li and lj, and at least one of these instructions
wrote Q.
1. li = read(Q), lj = read(Q). li and lj don’t conflict.
2. li = read(Q), lj = write(Q). They conflict.
3. li = write(Q), lj = read(Q). They conflict
4. li = write(Q), lj = write(Q). They conflict
 Intuitively, a conflict between li and lj forces an order between them.
 If li and lj are consecutive in a schedule and they do not conflict, their results
would remain the same even if they had been interchanged in the schedule.
Conflict Serializability
 If a schedule S can be transformed into a schedule S´ by a series of swaps of non-conflicting
instructions, we say that S and S´ are conflict equivalent.
 We say that a schedule S is conflict serializable if it is conflict equivalent to a serial schedule
 Schedule 3 can be transformed into Schedule 6, a serial schedule where T2 follows T1, by series
of swaps of non-conflicting instructions. Therefore it is conflict serializable.

Schedule 3 Schedule 6
Conflict Serializability (Cont.)
 Example of a schedule that is not conflict serializable:

 We are unable to swap instructions in the above schedule to obtain either the
serial schedule < T3, T4 >, or the serial schedule < T4, T3 >.
Testing for Serializability
 Consider some schedule of a set of transactions T1, T2, ..., Tn
 Precedence graph — a direct graph where
 the vertices are the transactions (names).
 there is an arc from Ti to Tj if the two transaction conflict, and Ti
accessed the data item on which the conflict arose earlier.
 We may label the arc by the item that was accessed.
 Example 1
x

y
Example Schedule (Schedule A) + Precedence Graph
T1 T2 T3 T4
T5
read(X)
read(Y)
read(Z)
read(V)
read(W) T T
read(W)
read(Y) 1 2
write(Y)
write(Z)
read(U)
read(Y)
write(Y)
read(Z) T T
write(Z)
4
read(U) 3
write(U)

T
5
Test for Conflict Serializability
 A schedule is conflict serializable if and only if its
precedence graph is acyclic.
 Cycle-detection algorithms exist which take order n2 time,
where n is the number of vertices in the graph.
 (Better algorithms take order n + e where e is the
number of edges.)
 If precedence graph is acyclic, the serializability order can
be obtained by a topological sorting of the graph.
 This is a linear order consistent with the partial order
of the graph.
 For example, a serializability order for Schedule A
would be
T5  T1  T3  T2  T4
View Serializability
 Sometimes it is possible to serialize schedules that are not conflict serializable
 View serializability provides a weaker and still consistency preserving notion of
serialization
 Let S and S´ be two schedules with the same set of transactions. S and S´ are
view equivalent if the following three conditions are met, for each data item
Q,
1. If in schedule S, transaction Ti reads the initial value of Q, then in schedule
S’ also transaction Ti must read the initial value of Q.
2. If in schedule S transaction Ti executes read(Q), and that value was
produced by transaction Tj (if any), then in schedule S’ also transaction Ti
must read the value of Q that was produced by the same write(Q) operation
of transaction Tj .
3. The transaction (if any) that performs the final write(Q) operation in
schedule S must also perform the final write(Q) operation in schedule S’.
Example: schedule 1 and schedule3 are view equivalent
View Serializability (Cont.)
 A schedule S is view serializable if it is view equivalent to a serial schedule.
 Every conflict serializable schedule is also view serializable.
 Below is a schedule which is view-serializable but not conflict serializable.

 It is equivalent to either <T3,T4,T6> or <T4,T3,T6>

 Every view serializable schedule that is not conflict serializable has blind
writes(write(Q) without performing read(Q)).
Test for View Serializability
 The precedence graph test for conflict serializability cannot be used directly to test
for view serializability.
 Extension to test for view serializability has cost exponential in the size of the
precedence graph.
 The problem of checking if a schedule is view serializable falls in the class of NP-
complete problems.
 Thus existence of an efficient algorithm is extremely unlikely.
 However practical algorithms that just check some sufficient conditions for view
serializability can still be used.
 Recoverable Schedules
What to do if some transaction fails? One needs to address
the effect of failures on concurrently running transactions.
 Recoverable schedule — if a transaction Tj reads a data item previously written by a
transaction Ti , then the commit operation of Ti must appear before the commit
operation of Tj.
 The following schedule is not recoverable if T9 commits immediately after the read

 If T8 should abort, T9 would have read (and possibly shown to the user) an inconsistent
database state. Hence, database must ensure that schedules are recoverable.
Cascading Rollbacks
 Cascading rollback – a single transaction failure leads to a series of transaction
rollbacks. Consider the following schedule where none of the transactions has
yet committed (so the schedule is recoverable)

If T10 fails, T11 and T12 must also be rolled back.

 Can lead to the undoing of a significant amount of work
Cascadeless Schedules

 Cascadeless schedules — in these, cascading rollbacks cannot occur; for each

pair of transactions Ti and Tj such that Tj reads a data item previously written
by Ti, the commit operation of Ti appears before the read operation of Tj.
 Every cascadeless schedule is also recoverable
 It is desirable to restrict the schedules to those that are cascadeless
Concurrency Control
 A database must provide a mechanism that will ensure that all possible schedules are
 either conflict or view serializable, and
 are recoverable and preferably cascadeless
 A policy in which only one transaction can execute at a time generates serial schedules,
but provides a poor degree of concurrency
 Are serial schedules recoverable/cascadeless?
 Testing a schedule for serializability after it has executed is a little too late!
 Goal – to develop concurrency control protocols that will assure serializability.
 Lock-based protocols
 Timestamp-based protocols
 Concurrency Control vs. Serializability Tests

 Concurrency-control protocols allow concurrent schedules, but ensure that the schedules are
conflict/view serializable, and are recoverable and cascadeless .
 Concurrency control protocols generally do not examine the precedence graph as it is being
created
 Instead a protocol imposes a discipline that avoids nonseralizable schedules.
 Different concurrency control protocols provide different tradeoffs between the amount of
concurrency they allow and the amount of overhead that they incur.
 Tests for serializability help us understand why a concurrency control protocol is correct.
Lock-Based Protocols
 A lock is a mechanism to control concurrent access to a data item
 Data items can be locked in two modes :
1. exclusive (X) mode. Data item 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.
 Lock requests are made to concurrency-control manager. Transaction can proceed only after request
is granted.
Lock-Based Protocols (Cont.)
 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.
Lock-Based Protocols (Cont.)
 Example of a transaction performing locking:
T2: lock-S(A);
read (A);
unlock(A);
lock-S(B);
read (B);
unlock(B);
display(A+B)
 Locking as above is not sufficient to guarantee serializability — if A and B get updated in-
between the read of A and B, the displayed sum would be wrong.
 A locking protocol is a set of rules followed by all transactions while requesting and
releasing locks. Locking protocols restrict the set of possible schedules.
The Two-Phase Locking Protocol
 This is a protocol which ensures conflict-serializable schedules.
 Phase 1: Growing Phase
 transaction may obtain locks
 transaction may not release locks
 Phase 2: Shrinking Phase
 transaction may release locks
 transaction may not obtain locks
 The 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).
Pitfalls of Lock-Based Protocols
 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.
 Such a situation is called a deadlock.
 To handle a deadlock one of T3 or T4 must be rolled back
and its locks released.
 Pitfalls of Lock-Based Protocols (Cont.)
 The potential for deadlock exists in most locking protocols. Deadlocks are a
necessary evil.
 Starvation is also possible if concurrency control manager is badly designed.
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.
 Concurrency control manager can be designed to prevent starvation.
 Two-phase locking does not ensure freedom from deadlocks
 Deadlock prevention protocols or deadlock detection mechanisms are
needed!
 With detection mechanisms 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.
Deadlock Detection
 Deadlocks can be described as a wait-for graph where:
 vertices are all the transactions in the system
 There is an edge Ti Tk in case Ti is waiting for Tk
 When Ti requests a data item currently being held by Tk, then the edge Ti Tk
is inserted in the wait-for graph. This edge is removed only when Tk is no
longer holding a data item needed by Ti.
 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.

Wait-for graph without a cycle Wait-for graph with a cycle

 Properties of the Two-Phase Locking Protocol
 Cascading 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.
 Rigorous two-phase locking is even stricter: here all locks are held till
commit/abort. In this protocol transactions can be serialized in the order in
which they commit.
 There can be conflict serializable schedules that cannot be obtained if two-
phase locking is used.
 However, 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 locking, and a schedule for Ti and Tj that
is not conflict serializable.
Multiple Granularity
 Up to now we have considered locking (and execution) at the level of a single item/row
 However there are circumstances at which it is preferable to perform lock at different
level (sets of tuples, relation, or even sets of relations)
 As extreme example consider a transaction that needs to access to whole database:
performing locks tuple by tuple would be time-consuming
 Allow data items to be of various sizes and define a hierarchy (tree) of data granularities,
where the small granularities are nested within larger ones
 When a transaction locks a node in the tree explicitly, it implicitly locks all the node's
descendents in the same mode.
 Granularity of locking (level in tree where locking is done):
 fine granularity (lower in tree): high concurrency, high locking overhead
 coarse granularity (higher in tree): low locking overhead, low concurrency
Example of Granularity Hierarchy

The levels, starting from the coarsest (top) level are

 database
 area
 file
 record
Timestamp-Based Protocols
 Instead of determining the order of each operation in a transaction at execution
time, determines the order by the time of beginning of each transaction.
 Each 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).
 The protocol manages concurrent execution such that the time-stamps
determine the serializability order.
 In order to assure such behavior, the protocol maintains for each data Q two
timestamp values:
 W-timestamp(Q) is the largest time-stamp of any transaction that executed
write(Q) successfully.
 R-timestamp(Q) is the largest time-stamp of any transaction that executed
read(Q) successfully.
 Timestamp-Based Protocols (Cont.)
 The timestamp ordering protocol ensures that any conflicting read and write
operations are executed in timestamp order.
 Suppose a transaction Ti issues a read(Q)
1. If TS(Ti)  W-timestamp(Q), then Ti needs to read a value of Q that was already
overwritten.
 Hence, 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)).
 Suppose 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.
 Hence, the write operation is rejected, and Ti is rolled back.
2. If TS(Ti) < W-timestamp(Q), then Ti is attempting to write an obsolete value of Q.
 Hence, 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).
 Correctness of Timestamp-Ordering Protocol

 The timestamp-ordering protocol guarantees serializability since all the arcs

in the precedence graph are of the form:

transaction transaction
with smaller with larger
timestamp timestamp

Thus, there will be no cycles in the precedence graph

 Timestamp protocol ensures freedom from deadlock as no transaction ever
waits.
 But the schedule may not be cascade-free, and may not even be
recoverable.
Multiversion Schemes
 Up to now we only considered a single copy (the most recent) of each database item.
 Multiversion schemes keep old versions of data item to increase concurrency.
 Multiversion Timestamp Ordering
 Multiversion Two-Phase Locking
 Basic Idea of multiversion schemes
 Each successful write results in the creation of a new version of the data item written.
 Use timestamps to label versions.
 When a read(Q) operation is issued, select an appropriate version of Q based on the timestamp
of the transaction, and return the value of the selected version.
 reads never have to wait as an appropriate version is returned immediately.
 A drawback is that creation of multiple versions increases storage overhead
 Garbage collection mechanism may be used…
Multiversion Timestamp Ordering
 Each data item Q has a sequence of versions <Q1, Q2,...., Qm>. Each version Qk contains
three data fields:
 Content -- the value of version Qk.
 W-timestamp(Qk) -- timestamp of the transaction that created (wrote) version Qk
 R-timestamp(Qk) -- largest timestamp of a transaction that successfully read
version Qk
 when a transaction Ti creates a new version Qk of Q, Qk's W-timestamp and R-
timestamp are initialized to TS(Ti).
 R-timestamp of Qk is updated whenever a transaction Tj reads Qk, and TS(Tj) > R-
timestamp(Qk).
Multiversion Timestamp Ordering (Cont)
 Suppose that transaction Ti issues a read(Q) or write(Q) operation. Let Qk denote the
version of Q whose write timestamp is the largest write timestamp less than or equal to
TS(Ti).
1. If transaction Ti issues a read(Q), then the value returned is the content of version
Qk.
2. If transaction Ti issues a write(Q)
1. if TS(Ti) < R-timestamp(Qk), then transaction Ti is rolled back.
2. if TS(Ti) = W-timestamp(Qk), the contents of Qk are overwritten
3. else a new version of Q is created.
 Observe that
 Reads always succeed
 A write by Ti is rejected if some other transaction Tj that (in the serialization order
defined by the timestamp values) should read
Ti's write, has already read a version created by a transaction older than Ti.
 Protocol guarantees serializability
Multiversion Two-Phase Locking
 Differentiates between read-only transactions and update transactions
 Update transactions acquire read and write locks, and hold all locks up to the end
of the transaction. That is, update transactions follow rigorous two-phase locking.
 Each successful write results in the creation of a new version of the data item
written.
 each version of a data item has a single timestamp whose value is obtained
from a counter ts-counter that is incremented during commit processing.
 Read-only transactions are assigned a timestamp by reading the current value of
ts-counter before they start execution; they follow the multiversion timestamp-
ordering protocol for performing reads.
Multiversion Two-Phase Locking (Cont.)
 When an update transaction wants to read a data item:
 it obtains a shared lock on it, and reads the latest version.
 When it wants to write an item
 it obtains X lock on; it then creates a new version of the item and sets this version's
timestamp to .
 When update transaction Ti completes, commit processing occurs:
 Ti sets timestamp on the versions it has created to ts-counter + 1
 Ti increments ts-counter by 1
 Read-only transactions that start after Ti increments ts-counter will see the values
updated by Ti.
 Read-only transactions that start before Ti increments the
ts-counter will see the value before the updates by Ti.
 Only serializable schedules are produced.
Weak Levels of Consistency

 Some applications are willing to live with weak levels of consistency, allowing
schedules that are not serializable
 E.g. a read-only transaction that wants to get an approximate total
balance of all accounts
 E.g. database statistics computed for query optimization can be
approximate
 Such transactions need not be serializable with respect to other
transactions
 Tradeoff accuracy for performance
Levels of Consistency in SQL-92
 Serializable — default in SQL standard
 Repeatable read — only committed records to be read, repeated reads of
same record must return same value (no updates of read items in between).
However, a transaction may not be serializable – it may find some records
inserted by a transaction but not find others.
 Read committed — only committed records can be read, but successive reads
of record may return different (but committed) values.
 Read uncommitted — even uncommitted records may be read.

 In many database systems, such as Oracle, read

committed is the default consistency level
 has to be explicitly changed to serializable when
required
 set isolation level serializable
 Lower degrees of consistency useful for gathering
approximate
information about the database
Recovery Schemes

 Recovery schemes are techniques to ensure database consistency and

transaction atomicity and durability despite failures such as transaction
failures, system crashes, disk failures.
 We just briefly focus this issue, which strongly relies on lower-level
control (usage of RAID, buffer management)
 More on this can be found in chapter 17 of the book
 Recovery algorithms have two parts
1. Actions taken during normal transaction processing to ensure enough
information exists to recover from failures
2. Actions taken after a failure to recover the database contents to a state
that ensures atomicity, consistency and durability
 Recovery and Atomicity
 Modifying the database without ensuring that the transaction commits may leave the
database in an inconsistent state.
 Consider again the transaction Ti that transfers €50 from account A to account B.
 Several output operations are required for Ti (to output A and B). A failure may
occur after one of these modifications have been made but before all of them are
made.
 To ensure atomicity despite failures, first output information describing the
modifications to stable storage (i.e. storage guaranteed/assumed not to fail, e.g.
with RAID) without modifying the database itself.
 Two approaches are possible:
 log-based recovery, and
 shadow-paging
Log-Based Recovery
 A log is kept on stable storage.
 The log is a sequence of log records, and maintains a record of update
activities on the database.
 When transaction Ti starts, it registers itself by writing a
<Ti start>log record
 Before Ti executes write(X), a log record <Ti, X, V1, V2> is written, where V1 is
the value of X before the write, and V2 is the value to be written to X.
 Log record notes that Ti has performed a write on data item Xj Xj had value
V1 before the write, and will have value V2 after the write.
 When Ti finishes it last statement, the log record <Ti commit> is written.
 For writing the actual records
 Deferred database modification
 Immediate database modification
Deferred Database Modification
 The deferred database modification scheme records all modifications to the
log, and defers writes to after partial commit.
 Transaction starts by writing <Ti start> record to log.
 A write(X) operation results in a log record <Ti, X, V> being written, where V
is the new value for X (old value is not needed).
 The write is not performed on X at this time, but is deferred.
 When Ti partially commits, <Ti commit> is written to the log
 After that, the log records are read and used to actually execute the previously
deferred writes.
 During recovery after a crash, a transaction needs to be redone if and only if
both <Ti start> and<Ti commit> are there in the log.
 Redoing a transaction Ti ( redoTi) sets the value of all data items updated by
the transaction to the new values.
Immediate Database Modification
 The immediate database modification scheme allows database updates of an uncommitted
transaction to be made as the writes are issued
 since undoing may be needed, update logs must have both old value and new value
 Update log record must be written before database item is written
 We assume that the log record is output directly to stable storage
 Can be extended to postpone log record output, so long as prior to execution of an
output(B) operation for a data block B, all log records corresponding to items B must be
flushed to stable storage
 Output of updated blocks can take place at any time before or after transaction commit
 Order in which blocks are output can be different from the order in which they are written.
Immediate Database Modification
(Cont.)
 Recovery procedure has two operations instead of one:
 undo(Ti) restores the value of all data items updated by Ti to their old values,
going backwards from the last log record for Ti
 redo(Ti) sets the value of all data items updated by Ti to the new values, going
forward from the first log record for Ti
 Both operations must be idempotent
 That is, even if the operation is executed multiple times the effect is the same
as if it is executed once
 Needed since operations may get re-executed during recovery
 When recovering after failure:
 Transaction Ti needs to be undone if the log contains the record
<Ti start>, but does not contain the record <Ti commit>.
 Transaction Ti needs to be redone if the log contains both the record <Ti start>
and the record <Ti commit>.
Checkpoints
 Problems in recovery procedure as discussed earlier :
1. searching the entire log is time-consuming
2. one might unnecessarily redo transactions which have already output their updates to
the database.
 Streamline recovery procedure by periodically performing checkpointing
1. Output all log records currently residing in main memory onto stable storage.
2. Output all modified buffer blocks to the disk.
3. Write a log record < checkpoint> onto stable storage.
Shadow Paging
 Shadow paging is an alternative to log-based recovery; this scheme is useful if
transactions execute serially
 Idea: maintain two page tables during the lifetime of a transaction –the
current page table, and the shadow page table
 Store the shadow page table in nonvolatile storage, such that state of the
database prior to transaction execution may be recovered.
 Shadow page table is never modified during execution
 To start with, both the page tables are identical. Only current page table is
used for data item accesses during execution of the transaction.
 Whenever any page is about to be written for the first time
 A copy of this page is made onto an unused page.
 The current page table is then made to point to the copy
 The update is performed on the copy
Transaction Definition in SQL
 Data manipulation language must include a construct for specifying the set of
actions that comprise a transaction.
 In SQL, a transaction begins implicitly, after previous transaction.
 A transaction in SQL ends by:
 Commit work commits current transaction and begins a new one.
 Rollback work causes current transaction to abort.
 In almost all database systems, by default, every SQL statement also commits
implicitly if it executes successfully
 Implicit commit can be turned off by a database directive
 E.g. in JDBC, connection.setAutoCommit(false);
 Four levels of (weak) consistency, cf. before.
Transaction management in Oracle

 Transaction beginning and ending as in SQL

 Explicit commit work and rollback work
 Implicit commit on session end, and implicit rollback on failure
 Log-based deferred recovery using rollback segment
 Checkpoints (inside transactions) can be handled explicitly
 savepoint <name>
 rollback to <name>

 Concurrency control is made by (a variant of) multiversion rigorous two-phase locking

 Deadlock are detected using a wait-graph
 Upon deadlock detection, the last transaction that detects the deadlock is rolled back
Levels of Consistency in Oracle

 Oracle implements 2 of the 4 of levels of SQL

 Read committed, by default in Oracle and with
 set transaction isolation level read committed
 Serializable, with
 set transaction isolation level serializable
 Appropriate for large databases with only few updates, and usually with
not many conflicts. Otherwise it is too costly.
 Further, it supports a level similar to repeatable read:
 Read only mode, only allow reads on committed data, and further doesn’t
allow INSERT, UPDATE or DELETE on that data. (without unrepeatable reads!)
 set transaction isolation level read only
Granularity in Oracle
 By default Oracle performs row level locking.
 Command
select … for update
locks the selected rows so that other users cannot lock or update the rows
until you end your transaction. Restriction:
 Only at top-level select (not in sub-queries)
 Not possible with DISTINCT operator, CURSOR expression, set operators,
group by clause, or aggregate functions.
 Explicit locking of tables is possible in several modes, with
 lock table <name> in
 row share mode
 row exclusive mode
 share mode
 share row exclusive mode
 exclusive mode
Lock modes in Oracle

 Row share mode

 The least restrictive mode (with highest degree of concurrency)
 Allows other transactions to query, insert, update, delete, or lock rows
concurrently in the same table, except for exclusive mode
 Row exclusive mode
 As before, but doesn’t allow setting other modes except for row share.
 Acquired automatically after a insert, update or delete command on a table
 Exclusive mode
 Only allows queries to records of the locked table
 No modifications are allowed
 No other transaction can lock the table in any other mode
Consistency tests in Oracle

 By default, in Oracle all consistency tests are made immediately after each DML command
(insert, delete or update).
 However, it is possible to defer consistency checking of constraints (primary keys,
candidate keys, foreign keys, and check conditions) to the end of transactions.
 Only this makes it possible e.g. to insert tuples in relation with circular dependencies
in foreign keys
 For this:
 each constraints that may possibly be deferred must be declared as deferrable:
 At the definition of the constraint add deferrable immediately afterwards
 at the transaction in which one wants to defer the verification of the constraints, add
command:
 set constraints all deferred
 In this command, instead of all it is possible to specify which constraints are to be
deferred, by giving their names separated by commas