...
| Info |
|---|
|
- Replicating DDL is not currently supported with CSV as a target.
- Adding (and preparing) new tables to the replication set is not currently supported when using Target CSV replication.
- In CSV output mode Dbvisit Replicate does its own buffering, and this is performed in memory. It essentially always operates according to a pessimistic commit model. This means that both the STAGING and TARGET directories only contain records of committed transactions, so you should not find the DATE_COMMIT column empty. However, at any point in time even for committed operations, some files are too small, or too early in the transaction process to be written.
- Ordering is preserved in transactions, and transaction order is also preserved. But transactions can overlap.
In other words - if the source has: - TX1 CH1
- TX2 CH2
- TX1 CH3
- TX2 CH4
Then we create the output such that: - if you look at TX1, you will see CH1, CH3
- if you look at TX2, you will see CH2, CH4
- if you look at TXs, you will see TX1, TX2.
But in reality you will see TX1 CH1, TX1 CH3, TX2 CH2, TX2 CH4. |
Setup Wizard Example
This example shows an Oracle-to-CSV one-way replication using the following information. We will be configuring the MINE and the APPLY on the same server in this example.
...