Pathway/iTS System Management Manual (G06.24+)

Tuning Your System Using Statistics

HP NonStop Pathway/iTS System Management Manual—426748-002

5-6

Checkpointing

Checkpointing refers to the communication between the primary and backup

processes of a TCP process pair.

During processing, the primary TCP keeps the backup TCP informed of what it is

doing—for example, sending a request, receiving a reply, and so on—with checkpoint

messages, as shown in Figure 5-1. Consequently, if the primary process fails, the

backup process has enough information to take over and continue.

Checkpointing occurs as follows:

1. A TCP task requests service by placing itself on the request queue associated with

checkpointing. While waiting for service, the TCP task cannot execute.

2. When the service completes, the TCP task is removed from the request queue.

If TMF is being used, the TCP checkpoints the task’s data area at the BEGIN

TRANSACTION and END TRANSACTION statements. If the TMF facility is not

installed, the TCP checkpoints before and after a SEND request to a server.

Gathering Statistics

A TCP gathers statistics about the tasks it performs, the terminals under its control,

and the servers it has open. These statistics are reported to the PATHMON process.

Thus, a TCP gathers TCP, TERM, and SERVER object statistics. Details about

statistics for TCP and TERM objects are included in separate subsections later in this

section. For information about SERVER object statistics, see the TS/MP System

Management Manual.

Figure 5-1. TCP Checkpointing

I'm functioning properly and

sending a request message to

Server XYZ.

Primary

TCP

Backup

TCP

I'm ready.

VST023.vsd