Download ch25

Document related concepts

Commitment ordering wikipedia , lookup

Serializability wikipedia , lookup

Concurrency control wikipedia , lookup

Transcript
Chapter 25
Distributed DBMSs - Advanced
Concepts
Pearson Education © 2009
Chapter 25 - Objectives
 Distributed
transaction management.
 Distributed concurrency control.
 Distributed deadlock detection.
 Distributed recovery control.
 Distributed integrity control.
 X/OPEN DTP standard.
 Distributed query optimization.
 Oracle’s DDBMS functionality.
Pearson Education © 2009
2
Distributed Transaction Management
 Distributed
transaction accesses data stored at
more than one location.
 Divided into a number of sub-transactions, one
for each site that has to be accessed, represented
by an agent.
 Indivisibility of distributed transaction is still
fundamental to transaction concept.
 DDBMS must also ensure indivisibility of each
sub-transaction.
Pearson Education © 2009
3
Distributed Transaction Management
 Thus,
DDBMS must ensure:
– synchronization of subtransactions with other
local transactions executing concurrently at a
site;
– synchronization of subtransactions with global
transactions running simultaneously at same
or different sites.
 Global
transaction manager (transaction
coordinator) at each site, to coordinate global
and local transactions initiated at that site.
Pearson Education © 2009
4
Coordination of Distributed Transaction
Pearson Education © 2009
5
Distributed Locking
 Look
–
–
–
–
at four schemes:
Centralized Locking.
Primary Copy 2PL.
Distributed 2PL.
Majority Locking.
Pearson Education © 2009
6
Centralized Locking
Single site that maintains all locking information.
 One lock manager for whole of DDBMS.
 Local transaction managers involved in global
transaction request and release locks from lock
manager.
 Or transaction coordinator can make all locking
requests on behalf of local transaction managers.
 Advantage - easy to implement.
 Disadvantages - bottlenecks and lower reliability.

Pearson Education © 2009
7
Primary Copy 2PL
 Lock
managers distributed to a number of sites.
 Each
lock manager responsible for managing
locks for set of data items.
 For
replicated data item, one copy is chosen as
primary copy, others are slave copies
 Only
need to write-lock primary copy of data item
that is to be updated.
 Once
primary copy has been updated, change can
be propagated to slaves.
Pearson Education © 2009
8
Primary Copy 2PL
 Disadvantages
- deadlock handling is more
complex; still a degree of centralization in
system.
 Advantages - lower communication costs and
better performance than centralized 2PL.
Pearson Education © 2009
9
Distributed 2PL
 Lock
managers distributed to every site.
 Each lock manager responsible for locks for
data at that site.
 If data not replicated, equivalent to primary
copy 2PL.
 Otherwise, implements a Read-One-Write-All
(ROWA) replica control protocol.
Pearson Education © 2009
10
Distributed 2PL
 Using
ROWA protocol:
– Any copy of replicated item can be used for
read.
– All copies must be write-locked before item
can be updated.
 Disadvantages - deadlock handling more
complex; communication costs higher than
primary copy 2PL.
Pearson Education © 2009
11
Majority Locking
Extension of distributed 2PL.
 To read or write data item replicated at n sites,
sends a lock request to more than half the n sites
where item is stored.
 Transaction cannot proceed until majority of
locks obtained.
 Overly strong in case of read locks.

Pearson Education © 2009
12
Distributed Timestamping
 Objective
is to order transactions globally so
older transactions (smaller timestamps) get
priority in event of conflict.
 In distributed environment, need to generate
unique timestamps both locally and globally.
 System clock or incremental event counter at
each site is unsuitable.
 Concatenate local timestamp with a unique site
identifier: <local timestamp, site identifier>.
Pearson Education © 2009
13
Distributed Timestamping
Site identifier placed in least significant position
to ensure events ordered according to their
occurrence as opposed to their location.
 To prevent a busy site generating larger
timestamps than slower sites:
– Each site includes their timestamps in messages.
– Site compares its timestamp with timestamp in
message and, if its timestamp is smaller, sets it
to some value greater than message timestamp.

Pearson Education © 2009
14
Distributed Deadlock
More complicated if lock management is not
centralized.
 Local Wait-for-Graph (LWFG) may not show
existence of deadlock.
 May need to create GWFG, union of all LWFGs.
 Look at three schemes:
– Centralized Deadlock Detection.
– Hierarchical Deadlock Detection.
– Distributed Deadlock Detection.

Pearson Education © 2009
15
Example - Distributed Deadlock



T1 initiated at site S1 and creating agent at S2,
T2 initiated at site S2 and creating agent at S3,
T3 initiated at site S3 and creating agent at S1.
Time
S1
S2
t1 read_lock(T1, x1) write_lock(T2, y2)
S3
read_lock(T3, z3)
t2 write_lock(T1, y1) write_lock(T2, z2)
t3 write_lock(T3, x1) write_lock(T1, y2)
Pearson Education © 2009
write_lock(T2, z3)
16
Example - Distributed Deadlock
Pearson Education © 2009
17
Centralized Deadlock Detection
 Single
site appointed deadlock detection
coordinator (DDC).
 DDC has responsibility for constructing and
maintaining GWFG.
 If one or more cycles exist, DDC must break
each cycle by selecting transactions to be rolled
back and restarted.
Pearson Education © 2009
18
Hierarchical Deadlock Detection
 Sites
are organized into a hierarchy.
 Each site sends its LWFG to detection site above
it in hierarchy.
 Reduces dependence on centralized detection
site.
Pearson Education © 2009
19
Hierarchical Deadlock Detection
Pearson Education © 2009
20
Distributed Deadlock Detection
Most well-known method developed by
Obermarck (1982).
 An external node, Text, is added to LWFG to
indicate remote agent.
 If a LWFG contains a cycle that does not involve
Text, then site and DDBMS are in deadlock.

Pearson Education © 2009
21
Distributed Deadlock Detection
 Global
deadlock may exist if LWFG contains a
cycle involving Text.
 To determine if there is deadlock, the graphs
have to be merged.
 Potentially more robust than other methods.
Pearson Education © 2009
22
Distributed Deadlock Detection
Pearson Education © 2009
23
Distributed Deadlock Detection
S1:
S2:
S3:
Text  T3  T1  Text
Text  T1  T2  Text
Text  T2  T3  Text
 Transmit
LWFG for S1 to the site for which
transaction T1 is waiting, site S2.
 LWFG at S2 is extended and becomes:
S2:
Text  T3  T1  T2  Text
Pearson Education © 2009
24
Distributed Deadlock Detection
 Still
contains potential deadlock, so transmit
this WFG to S3:
S3:
Text  T3  T1  T2  T3  Text
 GWFG
contains cycle not involving Text, so
deadlock exists.
Pearson Education © 2009
25
Distributed Deadlock Detection
 Four
types of failure particular to distributed
systems:
– Loss of a message.
– Failure of a communication link.
– Failure of a site.
– Network partitioning.
 Assume
first are handled transparently by DC
component.
Pearson Education © 2009
26
Distributed Recovery Control
 DDBMS
is highly dependent on ability of all
sites to be able to communicate reliably with
one another.
 Communication failures can result in network
becoming split into two or more partitions.
 May be difficult to distinguish whether
communication link or site has failed.
Pearson Education © 2009
27
Partitioning of a network
Pearson Education © 2009
28
Two-Phase Commit (2PC)
 Two
phases: a voting phase and a decision phase.
 Coordinator asks all participants whether they
are prepared to commit transaction.
– If one participant votes abort, or fails to
respond within a timeout period, coordinator
instructs all participants to abort transaction.
– If all vote commit, coordinator instructs all
participants to commit.
 All participants must adopt global decision.
Pearson Education © 2009
29
Two-Phase Commit (2PC)
If participant votes abort, free to abort
transaction immediately
 If participant votes commit, must wait for
coordinator to broadcast global-commit or
global-abort message.
 Protocol assumes each site has its own local log
and can rollback or commit transaction reliably.
 If participant fails to vote, abort is assumed.
 If participant gets no vote instruction from
coordinator, can abort.

Pearson Education © 2009
30
2PC Protocol for Participant Voting Commit
Pearson Education © 2009
31
2PC Protocol for Participant Voting Abort
Pearson Education © 2009
32
2PC Termination Protocols
 Invoked
whenever a coordinator or participant
fails to receive an expected message and times out.
Coordinator
 Timeout in WAITING state
– Globally abort transaction.
 Timeout
in DECIDED state
– Send global decision again to sites that have not
acknowledged.
Pearson Education © 2009
33
2PC - Termination Protocols (Participant)
 Simplest
termination protocol is to leave
participant blocked until communication with the
coordinator is re-established. Alternatively:
 Timeout
in INITIAL state
– Unilaterally abort transaction.
 Timeout
in the PREPARED state
– Without more information, participant blocked.
– Could get decision from another participant .
Pearson Education © 2009
34
State Transition Diagram for 2PC
(a) coordinator; (b) participant
Pearson Education © 2009
35
2PC Recovery Protocols
 Action
to be taken by operational site in event of
failure. Depends on what stage coordinator or
participant had reached.
Coordinator Failure
 Failure in INITIAL state
– Recovery starts commit procedure.
 Failure in WAITING state
– Recovery restarts commit procedure.
Pearson Education © 2009
36
2PC Recovery Protocols (Coordinator Failure)
 Failure
in DECIDED state
– On restart, if coordinator has received all
acknowledgements,
it
can
complete
successfully. Otherwise, has to initiate
termination protocol discussed above.
Pearson Education © 2009
37
2PC Recovery Protocols (Participant Failure)
 Objective
to ensure that participant on restart
performs same action as all other participants
and that this restart can be performed
independently.
 Failure
in INITIAL state
– Unilaterally abort transaction.
 Failure in PREPARED state
– Recovery via termination protocol above.
 Failure in ABORTED/COMMITTED states
– On restart, no further action is necessary.
Pearson Education © 2009
38
2PC Topologies
Pearson Education © 2009
39
Three-Phase Commit (3PC)
 2PC
is not a non-blocking protocol.
 For example, a process that times out after
voting commit, but before receiving global
instruction, is blocked if it can communicate only
with sites that do not know global decision.
 Probability of blocking occurring in practice is
sufficiently rare that most existing systems use
2PC.
Pearson Education © 2009
40
Three-Phase Commit (3PC)
 Alternative
non-blocking protocol, called threephase commit (3PC) protocol.
 Non-blocking for site failures, except in event of
failure of all sites.
 Communication failures can result in different
sites reaching different decisions, thereby
violating atomicity of global transactions.
 3PC removes uncertainty period for participants
who have voted commit and await global
decision.
Pearson Education © 2009
41
Three-Phase Commit (3PC)
 Introduces
third phase, called pre-commit,
between voting and global decision.
 On receiving all votes from participants,
coordinator sends global pre-commit message.
 Participant who receives global pre-commit,
knows all other participants have voted commit
and that, in time, participant itself will definitely
commit.
Pearson Education © 2009
42
State Transition Diagram for 3PC
(a) coordinator; (b) participant
Pearson Education © 2009
43
3PC Protocol for Participant Voting Commit
44
Pearson Education © 2009
3PC Termination Protocols (Coordinator)
 Timeout
in WAITING state
– Same as 2PC. Globally abort transaction.
 Timeout
in PRE-COMMITTED state
– Write commit record to log and send
GLOBAL-COMMIT message.
 Timeout in DECIDED state
– Same as 2PC. Send global decision again to
sites that have not acknowledged.
Pearson Education © 2009
45
3PC Termination Protocols (Participant)
 Timeout
in INITIAL state
– Same as 2PC. Unilaterally abort transaction.
 Timeout in the PREPARED state
– Follow election protocol to elect new coordinator.
 Timeout in the PRE-COMMITTED state
– Follow election protocol to elect new coordinator.
Pearson Education © 2009
46
3PC Recovery Protocols (Coordinator Failure)
 Failure
in INITIAL state
– Recovery starts commit procedure.
 Failure
in WAITING state
– Contact other sites to determine fate of transaction.
 Failure
in PRE-COMMITTED state
– Contact other sites to determine fate of transaction.
 Failure
in DECIDED state
– If all acknowledgements in, complete transaction;
otherwise initiate termination protocol above.
Pearson Education © 2009
47
3PC Recovery Protocols (Participant Failure)
 Failure
in INITIAL state
– Unilaterally abort transaction.
 Failure in PREPARED state
– Contact other sites to determine fate of
transaction.
 Failure in PRE-COMMITTED state
– Contact other sites to determine fate of
transaction.
 Failure in ABORTED/COMMITTED states
– On restart, no further action is necessary.
Pearson Education © 2009
48
3PC Termination Protocol After New Coordinator

1.
2.
3.
4.
Newly elected coordinator will send STATEREQ message to all participants involved in
election to determine how best to continue.
If some participant has aborted, then abort.
If some participant has committed, then
commit.
If all participants are uncertain, then abort.
If some participant is in PRE-COMMIT, then
commit. To prevent blocking, send PRECOMMIT and after acknowledgements, send
GLOBAL-COMMIT.
Pearson Education © 2009
49
Network Partitioning
 If
data is not replicated, can allow transaction to
proceed if it does not require any data from site
outside partition in which it is initiated.
 Otherwise, transaction must wait until sites it
needs access to are available.
 If data is replicated, procedure is much more
complicated.
Pearson Education © 2009
50
Identifying Updates
Pearson Education © 2009
51
Identifying Updates
 Successfully
completed update operations by
users in different partitions can be difficult to
observe.
 In P1, transaction withdrawn £10 from account
and in P2, two transactions have each withdrawn
£5 from same account.
 At start, both partitions have £100 in balx, and on
completion both have £90 in balx.
 On recovery, not sufficient to check value in balx
and assume consistency if values same.
Pearson Education © 2009
52
Maintaining Integrity
Pearson Education © 2009
53
Maintaining Integrity
Successfully completed update operations by users
in different partitions can violate constraints.
 Have constraint that account cannot go below £0.
 In P1, withdrawn £60 from account and in P2,
withdrawn £50.
 At start, both partitions have £100 in balx, then on
completion one has £40 in balx and other has £50.
 Importantly, neither has violated constraint.
 On recovery, balx is –£10, and constraint violated.

Pearson Education © 2009
54
Network Partitioning
 Processing
in partitioned network involves tradeoff in availability and correctness.
 Correctness easiest to provide if no processing of
replicated data allowed during partitioning.
 Availability maximized if no restrictions placed
on processing of replicated data.
 In general, not possible to design non-blocking
commit protocol for arbitrarily partitioned
networks.
Pearson Education © 2009
55
X/OPEN DTP Model
 Open
Group is vendor-neutral consortium whose
mission is to cause creation of viable, global
information infrastructure.
 Formed by merge of X/Open and Open Software
Foundation.
 X/Open established DTP Working Group with
objective of specifying and fostering appropriate
APIs for TP.
 Group concentrated on elements of TP system
that provided the ACID properties.
Pearson Education © 2009
56
X/OPEN DTP Model
 X/Open
DTP standard that emerged specified
three interacting components:
– an application,
– a transaction manager (TM),
– a resource manager (RM).
Pearson Education © 2009
57
X/OPEN DTP Model
Any subsystem that implements transactional data
can be a RM, such as DBMS, transactional file
system or session manager.
 TM responsible for defining scope of transaction,
and for assigning unique ID to it.
 Application calls TM to start transaction, calls
RMs to manipulate data, and calls TM to
terminate transaction.
 TM communicates with RMs to coordinate
transaction, and TMs to coordinate distributed
transactions.

Pearson Education © 2009
58
X/OPEN DTP Model - Interfaces
 Application
may use TX interface to
communicate with a TM.
 TX provides calls that define transaction scope,
and whether to commit/abort transaction.
 TM communicates transactional information
with RMs through XA interface.
 Finally, application can communicate directly
with RMs through a native API, such as SQL or
ISAM.
Pearson Education © 2009
59
X/OPEN DTP Model Interfaces
Pearson Education © 2009
60
X/OPEN Interfaces in Distributed Environment
Pearson Education © 2009
61
Distributed Query Optimization
62
Pearson Education © 2009
Distributed Query Optimization
Query decomposition: takes query expressed on
global
relations
and
performs
partial
optimization using centralized QO techniques.
Output is some form of RAT based on global
relations.
 Data localization: takes into account how data
has been distributed. Replace global relations at
leaves of RAT with their reconstruction
algorithms.

Pearson Education © 2009
63
Distributed Query Optimization
Global optimization: uses statistical information
to find a near-optimal execution plan. Output is
execution strategy based on fragments with
communication primitives added.
 Local optimization: Each local DBMS performs
its own local optimization using centralized QO
techniques.

Pearson Education © 2009
64
Data Localization
In QP, represent query as R.A.T. and, using
transformation rules, restructure tree into
equivalent form that improves processing.
 In DQP, need to consider data distribution.
 Replace global relations at leaves of tree with
their reconstruction algorithms - RA operations
that reconstruct global relations from fragments:

– For
horizontal
fragmentation,
algorithm is Union;
– For vertical fragmentation, it is Join.
Pearson Education © 2009
reconstruction
65
Data Localization
 Then
use reduction techniques to generate
simpler and optimized query.
 Consider reduction techniques for following types
of fragmentation:
– Primary horizontal fragmentation.
– Vertical fragmentation.
– Derived fragmentation.
Pearson Education © 2009
66
Reduction for Primary Horizontal Fragmentation
 If
selection predicate contradicts definition of
fragment, this produces empty intermediate
relation and operations can be eliminated.
 For join, commute join with union.
 Then
examine each individual join to determine
whether there are any useless joins that can be
eliminated from result.
 A useless
join exists if fragment predicates do not
overlap.
Pearson Education © 2009
67
Example 25.2 Reduction for PHF
SELECT *
FROM Branch b, PropertyForRent p
WHERE b.branchNo = p.branchNo AND p.type = ‘Flat’;
P1:
branchNo=‘B003’  type=‘House’ (PropertyForRent)
P2:
branchNo=‘B003’  type=‘Flat’ (PropertyForRent)
P3:
branchNo!=‘B003’ (PropertyForRent)
B1:
branchNo=‘B003’ (Branch)
B2:
branchNo!=‘B003’ (Branch)
Pearson Education © 2009
68
Example 25.2 Reduction for PHF
Pearson Education © 2009
69
Example 25.2 Reduction for PHF
Pearson Education © 2009
70
Example 25.2 Reduction for PHF
Pearson Education © 2009
71
Reduction for Vertical Fragmentation
 Reduction
for vertical fragmentation involves
removing those vertical fragments that have no
attributes in common with projection
attributes, except the key of the relation.
Pearson Education © 2009
72
Example 25.3 Reduction for Vertical Fragmentation
SELECT fName, lName
FROM Staff;
S1:
staffNo, position, sex, DOB, salary(Staff)
S2:
staffNo, fName, lName, branchNo (Staff)
Pearson Education © 2009
73
Example 25.3 Reduction for Vertical Fragmentation
Pearson Education © 2009
74
Reduction for Derived Fragmentation
 Use
transformation rule that allows join and
union to be commuted.
 Using knowledge that fragmentation for one
relation is based on the other and, in
commuting, some of the partial joins should be
redundant.
Pearson Education © 2009
75
Example 25.4 Reduction for Derived Fragmentation
SELECT *
FROM Branch b, Client c
WHERE b.branchNo = c.branchNo AND
b.branchNo = ‘B003’;
B1 = branchNo=‘B003’ (Branch)
B2 = branchNo!=‘B003’ (Branch)
Ci = Client branchNo Bi
i = 1, 2
Pearson Education © 2009
76
Example 25.4 Reduction for Derived Fragmentation
Pearson Education © 2009
77
Global Optimization
Objective of this layer is to take the reduced
query plan for the data localization layer and
find a near-optimal execution strategy.
 In distributed environment, speed of network has
to be considered when comparing strategies.
 If know topology is that of WAN, could ignore all
costs other than network costs.
 LAN typically much faster than WAN, but still
slower than disk access.

Pearson Education © 2009
78
Global Optimization
 Cost
model could be based on total cost (time),
as in centralized DBMS, or response time.
Latter uses parallelism inherent in DDBMS.
Pearson Education © 2009
79
Global Optimization – R*
R* uses a cost model based on total cost and
static query optimization.
 Like centralized System R optimizer, algorithm is
based on an exhaustive search of all join
orderings, join methods (nested loop or sortmerge join), and various access paths for each
relation.
 When Join is required involving relations at
different sites, R* selects the sites to perform Join
and method of transferring data between sites.

Pearson Education © 2009
80
Global Optimization – R*

For a Join of R and S with R at site 1 and S at site
2, there are three candidate sites:
– site 1, where R is located;
– site 2, where S is located;
– some other site (e.g., site of relation T, which is
to be joined with join of R and S).
Pearson Education © 2009
81
Global Optimization – R*

In R*, there are 2 methods for transferring data:
1. Ship whole relation
2. Fetch tuples as needed.


First method incurs a larger data transfer but
fewer message then second.
R* considers only the following methods:
1. Nested loop, ship whole outer relation to site of inner.
2. Sort-merge, ship whole inner relation to site of outer.
3. Nested loop, fetch tuples of inner relation as needed
for each tuple of outer relation.
4. Sort-merge, fetch tuples of inner relation as needed for
each tuple of outer relation.
5. Ship both relations to third site.
Pearson Education © 2009
82
Global Optimization – SDD-1
Based on an earlier method known as “hill
climbing”, a greedy algorithm that starts with an
initial feasible solution which is then iteratively
improved.
 Modified to make use of Semijoin to reduce
cardinality of join operands.
 Like R*, SDD-1 optimizer minimizes total cost,
although unlike R* it ignores local processing
costs and concentrates on communication
message size.
 Like R*, query processing timing used is static.

Pearson Education © 2009
83
Global Optimization – SDD-1
Based on concept of “beneficial Semijoins”.
 Communication cost of Semijoin is simply cost of
transferring join attribute of first operand to site
of second operand.
 “Benefit” of Semijoin is taken as cost of
transferring irrelevant tuples of first operand,
which Semijoin avoids.

Pearson Education © 2009
84
Global Optimization – SDD-1
Phase 1 – Initialization: Perform all local
reductions using Selection and Projection.
Execute Semijoins within same site to reduce
sizes of relations. Generate set of all beneficial
Semijoins across sites (Semijoin is beneficial if its
cost is less than its benefit).
 Phase 2 – Selection of beneficial Semijoins:
Iteratively select most beneficial Semijoin from
set generated and add it to execution strategy.
After each iteration, update database statistics to
reflect incorporation of the Semijoin and update
the set with new beneficial Semijoins.

Pearson Education © 2009
85
Global Optimization – SDD-1
Phase 3 – Assembly site selection: Select, among all
sites, site to which transmission of all relations incurs
a minimum cost. Choose site containing largest
amount of data after reduction phase so that sum of
the amount of data transferred from other sites will
be minimum.
 Phase 4 – Postoptimization: Discard useless
Semijoins; e.g. if R resides in assembly site and R is
due to be reduced by Semijoin, but is not used to
reduce other relations after Semijoin, then since R
need not be moved to another site during assembly
phase, Semijoin on R is useless and can be discarded.

Pearson Education © 2009
86
Oracle’s DDBMS Functionality
 Oracle does not support type of fragmentation
discussed previously, although DBA can
distribute data to achieve similar effect.
 Thus,
fragmentation transparency is not
supported although location transparency is.
 Discuss:
–
–
–
–
–
–
connectivity
global database names and database links
transactions
referential integrity
heterogeneous distributed databases
Distributed QO.
Pearson Education © 2009
87
Connectivity – Oracle Net Services
 Oracle
Net Services supports communication
between clients and servers.
 Enables both client-server and server-server
communication across any network, supporting
both distributed processing and distributed
DBMS capability.
 Also responsible for translating any differences in
character sets or data representation that may
exist at operating system level.
Pearson Education © 2009
88
Global Database Names
 Unique
name given to each distributed database.
 Formed by prefixing the database’s network
domain name with the local database name.
 Domain
name follows standard Internet
conventions, with levels separated by dots
ordered from leaf to root, left to right.
89
Pearson Education © 2009
Database Links
 Used
to build distributed databases.
 Defines a communication path from one Oracle
database to another (possibly non-Oracle)
database.
 Acts as a type of remote login to remote database.
CREATE PUBLIC DATABASE LINK
RENTALS.GLASGOW.NORTH.COM;
SELECT *
FROM [email protected];
UPDATE [email protected]
SET salary = salary*1.05;
Pearson Education © 2009
90
Types of Transactions
 Remote
SQL statements: Remote query selects
data from one or more remote tables, all of which
reside at same remote node. Remote update
modifies data in one or more tables, all of which
are located at same remote node .
 Distributed SQL statements: Distributed query
retrieves data from two or more nodes.
Distributed update modifies data on two or more
nodes.
 Remote transactions: Contains one or more
remote statements, all of which reference a single
remote node.
Pearson Education © 2009
91
Types of Transactions
 Distributed
transactions: Includes one or more
statements that, individually or as a group,
update data on two or more distinct nodes of a
distributed database. Oracle ensures integrity of
distributed transactions using 2PC.
Pearson Education © 2009
92
Referential Integrity
 Oracle
does not permit declarative referential
integrity constraints to be defined across
databases.
 However, parent-child table relationships across
databases can be maintained using triggers.
Pearson Education © 2009
93
Heterogeneous Distributed Databases
 Here
one of the local DBMSs is not Oracle.
 Oracle Heterogeneous Services and a non-Oracle
system-specific agent can hide distribution and
heterogeneity.
 Can be accessed through:
– transparent gateways
– generic connectivity.
Pearson Education © 2009
94
Transparent Gateways
95
Pearson Education © 2009
Generic Connectivity
96
Pearson Education © 2009
Oracle Distributed Query Optimization
A
distributed query is decomposed by the local
Oracle DBMS into a number of remote queries,
which are sent to remote DBMS for execution.
 Remote DBMSs execute queries and send results
back to local node.
 Local node then performs any necessary
postprocessing and returns results to user.
 Only necessary data from remote tables are
extracted, thereby reducing amount of data that
needs to be transferred.
Pearson Education © 2009
97