Determining the size of trail file areas
You would most likely create an Extract process with a dedicated Datapump process in the source environment. The Datapump process can suffer an outage in case of any network issue. However, it, resumes the trail file transfer to the remote site when it is up again. In such a situation you would want the Extract process to continue extracting the data into the trail files on the source system.
The trail files are stored in the filesystem which would be finite in space. The size of the trail file area will determine how many trail files can be stored, which in effect will determine how long the Extract process can continue generating the trail files while the Datapump process is down. This will determine the overall network outage capacity of your replication infrastructure. Clearly, it is quite important to size the trail file areas appropriately to avoid the situations where the Extract process has to be stopped because of lack of available space on the filesystem.
In this recipe we will go through the method which you can use to calculate the required trail file area sizes for your environment.
How to do it…
- Calculate the trail file area size using the following formula:
Trail file area size = Redo Log Volume in an hour X Maximum number of hours downtime X .4
How it works...
For this formula, you need to determine the amount of redo logs generated in the busiest period for your system and the maximum amount of replication downtime that you can bear for your target system. Oracle recommends multiplying the product of these with .4 as usually GoldenGate requires only 40 percent of the data from the archive logs.
You should however, test the size of trail files generated for a period and measure it with the size of the archive files generated to determine the correct ratio for your system.
Tip
Since GoldenGate stores Object Names with each update in trail files, if the names of the objects in your system are quite long, you may find that the ratio of trail file size to archive log size is much higher than 4:10.
General rule of thumb for sizing the trail files is keeping them same as the Archive Log size. Keeping it too small will cause very frequent switches which will cause performance overhead.