How it works...

In larger implementation systems, where large volume of data is being replicated, there will likely be more parallel processes running in both the source and target sites to improve data throughput in your environment. Oracle GoldenGate can support up to 5,000 concurrent Extract and Replicat processes per instance. Each Extract and Replicat process needs a minimum of 20-55 MB of memory. Memory can grow per process depending on the size and mix of concurrent transactions.

Oracle GoldenGate memory consumption is only restrained by the physical memory managed by the operating system, and not the Oracle GoldenGate program. However, in recent releases, Oracle GoldenGate cache manager takes advantage of the operating system's memory management functions, thus making more efficient use of memory.

It is advisable to create a separate filesystem for your OGG installation software as follows:

If you are installing on a cluster environment, such as Oracle RAC, ensure the software distribution is on a shared filesystem so that each instance of your RAC database can access the binaries.

An additional 1 GB of disk space is required to hold your OGG trail files. This is an estimate and your environment may need more or less, depending of the amount of data volume to process and also factoring in the retention period of these trail files. I've seen sites where trail files are kept for seven days before being purged, or for one to two days. Therefore, the retention period will add to your storage requirements.

Regarding ping stats: ping was issued with a count of 0.8 Round Trip Time (RTT) to gauge for any latency fluctuation in the network. The RTT 0.8 ms is then multiplied by the network bandwidth (for example, your interface card speed). We divide the result by 8 to determine the number of bytes (8 bits = 1 byte). This value yields 1,000,000 bytes. Therefore the data pump extract parameter should reflect the following: tcpbufsize 1000000, for your optimum transfer rate.

Use the compress option of the RMTHOST parameter to compress data before it is sent over the wire. You would normally consider this given that your enterprise may have a very busy network, or if your network is very slow due to low bandwidth throughput. Make sure you weigh in the benefits of compression against the CPU resources that are required to perform the compression.