Options for loading target tables depend on the type of
Q Replication that you are setting up.
The following sections explain what load options are available
for each type of Q Replication, which server is used for the initial
load, and the steps that you must take to begin the load:
Load options for unidirectional replication
This
section explains the load options for unidirectional replication.
- Load options
- The available options depend on the target type:
- DB2® targets
- All load options are available.
- CCD targets
- If the CCD table is noncomplete, the only valid choice is no load.
If the CCD table is complete, all load options are available.
- Non-DB2 targets
- For automatic loads, the EXPORT and IMPORT utilities are supported.You can also specify that Q Apply
use an ODBC select to fetch data from the source, and for Oracle targets,
you can specify that Q Apply call the SQL*Loader utility.
- Which server is used for the initial load
- Q Capture server
- What you must do
- By default, the Q Apply program begins the loading process when
you start the Q Capture program for the Q subscription's source table.
If you create the Q subscription so that it is not started automatically
when the Q Capture program starts, then you must start the Q subscription
for the load to begin.
- Example
Assume that you want to replicate data in one direction from
the DEPARTMENT table at Server A to the DEPARTMENT table at Server
B and use the most automatic method. You want the Q Apply program
to handle the load and use the best available utility.
- You create a Q subscription for the DEPARTMENT table that specifies
an automatic load that uses the best available utility.
- You start the Q Capture program at Server A and the Q Apply program
at Server B.
The Q Apply program calls a load utility that copies
the data from the DEPARTMENT table at Server A to the DEPARTMENT table
at Server B. Once the loading process is finished, replication begins
in one direction from Server A to Server B.
Load options for bidirectional and peer-to-peer
replication with two servers
This section explains the load
options for bidirectional and peer-to-peer replication with two servers.
- Load options
- All load options are available. However, if you specify an automatic
load, by default the Q Apply program will choose between a combination
of the EXPORT and LOAD utilities and a combination of the EXPORT and
IMPORT utilities, depending on your operating system and version.
You can override this behavior and instruct the Q Apply program to
use the LOAD from CURSOR utility by opening the Q Subscription Properties
notebook for individual Q subscriptions.
- Which server is used for the initial load
- When you create the two Q subscriptions for bidirectional or peer-to-peer
replication with two servers, you choose which server will be the
initial load source. This server contains the table whose data you
want to copy to a table on the other server.
For subsequent loads
(for example, if you stop the Q subscriptions for the logical table
and then start them), you specify which server will be the load source
when you decide which of the two Q subscriptions to start. The source
table for the Q subscription that you start will be the load source.
- What you must do
- The process of initiating a load differs depending on whether
you specify an automatic or manual load:
- Automatic load
- If you created the Q subscriptions to start automatically when
the Q Capture program starts, you only need to start the Q Capture
and Q Apply programs at both servers for the loading process to begin.
If
you chose not to have the Q subscriptions start automatically, you
must take the following actions:
- Start the Q Capture and Q Apply programs at both servers.
- Start the Q subscription whose source table you specified as the
load source.
- Manual load
- Start the Q Capture and Q Apply programs at both servers.
- Start the Q subscription whose source table you want to be the
load source.
The Q subscription will go into load pending state.
- Load the target table for the Q subscription, using any method.
- When you are done with the load, tell the Replication Center that
the load is finished or insert a LOADDONE signal into the IBMQREP_SIGNAL
table at the source server for the Q subscription.
- Example
Assume that you wanted to replicate the EMPLOYEE table in a
bidirectional setup on Server A and Server B, and use the most automatic
method. You want Server A to be the initial load source:
- You create two Q subscriptions, EMP_A2B and EMP_B2A. When you
create EMP_A2B, you specify Server A as the initial load source and
specify an automatic load in which the Q Apply program chooses the
best available load utility.
- You initiate the load by starting the Q Capture and Q Apply programs
at Server A and Server B.
The Q Apply program at Server B initiates
the load for EMP_A2B by calling a load utility to copy the data from
the EMPLOYEE table at Server A to the EMPLOYEE table at Server B.
When the loading completes, replication begins in both directions
between Server A and Server B.
Load options for peer-to-peer replication
with three or more servers
This section explains the load
options for peer-to-peer replication with three or more servers.
- Load options
- All load options are available. However, if you specify an automatic
load, by default the Q Apply program chooses between a combination
of the EXPORT and LOAD utilities, and a combination of the EXPORT
and IMPORT utilities, depending on your operating system and version.
You can override this behavior and instruct the Q Apply program to
use the LOAD from CURSOR utility by opening the Q Subscription Properties
notebook for individual Q subscriptions.
- Which server is used for the initial load
- In a peer-to-peer group with three or more servers, you start
replication in stages. First you start replication between two servers,
and then you bring additional servers into the group by starting replication
between an active server and a new server. Follow these guidelines:
- When you start replication between the first two servers, choose
one server as the load source. Start the Q subscription that specifies
this server as its source. The Q Apply program at the second server
begins the loading process for the table at the second server.
- To add a new server, choose one of the active servers as the load
source. Start the Q subscription that specifies this server as its
source and the new server as its target. The Q Apply program at the
new server begins the loading process for the table at the new server.
- What you must do
- In a peer-to-peer configuration with three or more servers, you
cannot create Q subscriptions that start automatically. You must manually
start the Q subscriptions in stages. Follow these steps:
- Start the Q Capture and Q Apply programs at the first two servers
in the group.
- Start one of the two Q subscriptions between the servers. The
source table for the Q subscription that you start will be the load
source, and the target table will be loaded.
- Start the Q Capture and Q Apply programs at a new server.
- Start a Q subscription that specifies one of the active servers
as its source, and the new server as its target. The source table
for the Q subscription that you start will be the load source, and
the table at the new server will be loaded.
- Follow Steps 3 and 4 until all the servers in the group are loaded.
Manual load: If you choose a manual load, you must
load the target table after you start each Q subscription, and then
notify the replication programs when the target table is loaded.
- Example
Assume that you want to initiate the loading process for a
peer-to-peer Q subscription group that includes Server A, Server B,
and Server C, with a single logical table, the DEPARTMENT table. You
want the Q Apply program to handle the loading and use the best available
load utility. You will use Server A as the load source for the tables
at both Server B and Server C.
- You create six Q subscriptions, DEP_A2B, DEP_B2A, DEP_A2C, DEP_C2A,
DEP_B2C, and DEP_C2B, all specifying an automatic load using the best
available utility.
- You start the Q Capture and Q Apply programs at Server A and Server
B.
- You start the Q subscription DEP_A2B.
The Q Apply program at
Server B calls a utility to load the DEPARTMENT table at Server B
with data from the DEPARTMENT table at Server A. When the loading
completes, replication begins in both directions between Server A
and Server B.
- To begin the load at Server C, you first start the Q Capture and
Q Apply programs at Server C.
- Next, you start the Q subscription DEP_A2C.
The Q Apply program
at Server C calls a utility to load the DEPARTMENT table at Server
C with data from the DEPARTMENTS table at Server A. When the loading
completes, replication begins in all directions between all three
servers.
Q Replication and Event Publishing