Special Considerations when using zFBA with IBM Connect:Direct
- Both Connect:Direct for z/OS nodes MUST run in separate LPARS
- Connect:Direct for z/OS nodes require the same zFBA configuration defined in both NETMAP files
The only protocols that support zFBA transfers are TCPIP and UDT; however, it is highly recommend that TCPIP be used. Any other protocol used will generate message SCZF002I and the COPY step will continue without using zFBA.
Once the data path is determined to be zFBA, if an allocation error occurs on z/OS or UNIX, the connection terminates and message SCZF003E or SCZF006E is generated. SCZF006E is a special case indicating that the PNODE and SNODE are on the same LPAR. SCZF003E errors are repeatable.
IBM® Connect:Direct uses a fixed size of 32MB regardless of the actual zFBA device size; therefore, the minimum device size is 32MB. Each RU is rounded up to the next zFBA block size multiple (for the current generation DS8K, there are 512 bytes per block). As many whole RUs are packed in a 32MB chunk as will fit.
CDPLEX supports zFBA transfers.
You can not modify the zFBA setting using the INITPARM refresh. You must utilize the new global initialization parameter of zFBA so the WLM can direct the process to the correct server that supports zFBA.
The following is not supported when using zFBA: Secure+, CTCA, and any SNA protocol.