Checkpoint/Restart Examples for TCP/IP or LU6.2 Transfers

The following examples describe how checkpoint/restart works for different transfer types.

TCP/IP or LU6.2 Transfers

In the case of transfers using TCP/IP IBM® Connect:Direct® sends approximately 30 bytes of overhead after sending an amount of data equal to the checkpoint interval in effect. Therefore, if the file size is 3 million bytes and the checkpoint interval is 10k, then 30 bytes of overhead is sent 300 times, resulting in a total overhead of 9000 bytes.

Other Transfers

For other transfers (LU0 or transfers to MS-DOS), IBM Connect:Direct sends an additional seven bytes per block or ten bytes per record of overhead during data transfer. Data is sent in record mode when reblocking is taking place during transfer (source and destination block sizes differ). When no reblocking is taking place, IBM Connect:Direct sends in block mode.

When you enable checkpointing, IBM Connect:Direct transfers positioning information in addition to the file. The positioning information enables IBM Connect:Direct to reposition the file in the event of an interruption. The following examples show how to determine this overhead in number of bytes when using Checkpoint/Restart for non-TCP/IP transfers.

In the following table, IBM Connect:Direct checkpoints the file in block mode and adds seven bytes to each block transmitted. This functionality adds only 2,100 bytes to the transmission of almost 6 million bytes.

File Attributes Sending File Receiving File
BLKSIZE 19,069 19,069
LRECL 0 0
DSORG PS PS
RECFM U U
FILE SIZE 300 blocks 300 blocks

In the following table, IBM Connect:Direct checkpoints the file in record mode because the data is being reblocked. IBM Connect:Direct adds 10 bytes to each record being transmitted. This functionality adds 150,000 bytes to the transmission, or 5 percent overall.

File Attributes Sending File Receiving File
BLKSIZE 5,000 2,000
LRECL 200 200
DSORG PS PS
RECFM FB FB
FILE SIZE 600 blocks 1,500 blocks