Moving Forward with the On Demand Real-time Enterprise

226 Moving Forward with the On Demand Real-time Enterprise

Figure 5-1 SQL Replication

5.1.2 Q-replication

Q replication is a solution that can replicate large volumes of data at very low

levels of latency. Q replication captures changes to source tables and converts

committed transactional data to messages. This data is put into queues as soon

as it is committed at the source, and then read by Q replication. This technology

is different from SQL replication. For example, the data is not staged in tables. In

this process, the data is put in messages that are sent to the target location

through WebSphere MQ message queues. Here, the messages are read from

the queues and then converted back into transactional data. The transactions

are then applied to the target tables with a highly parallel method that is very fast,

and still preserves the integrity of the data.

You can replicate a subset of columns and rows from the source tables. All

subsetting occurs at the source location so that only the data you want is

transported across the network. If you want to perform data transformations, you

can pass replicated data to your own stored procedures.

This process is depicted in Figure 5-2. Here the committed transaction data is

read from the DB2 log tables by the capture program. However, the data is then

put into WebSphere message queues rather than into DB2 tables. The apply

program is activated when there is a message in the queue. It reads the