In that figure, two update transactions are performed. The first update goes against
a local table (EMPLOYEES); the second, against a remote table (EMPLOYEES@HR_LINK). After
FIGURE 17-2 Sample remote query
FIGURE 17-3 Sample distributed transaction
Chapter 17: Managing Distributed Databases 603
the two transactions are performed, a single commit is then executed. If either transaction cannot
commit, both transactions will be rolled back.
Distributed transactions yield two important benefits: databases on other servers can be
updated, and those transactions can be grouped together with others in a logical unit. The second
benefit occurs because of the database??™s use of 2PC. Here are the two phases:
The prepare phase An initiating node called the global coordinator notifies all sites
involved in the transaction to be ready either to commit or to roll back the transaction.
The commit phase If there is no problem with the prepare phase, all sites commit their
transactions. If a network or node failure occurs, all sites roll back their transactions.
The use of 2PC is transparent to the users. If the node that initiates the transaction forgets
about the transaction, a third phase, the forget phase, is performed. The detailed management of
distributed transactions is discussed in the section ???Managing Distributed Transactions,??? later in
this chapter.
Dynamic Data Replication
To improve the performance of queries that use data from remote databases, you may wish to
replicate that data on the local server.
Pages:
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884