OS/390 Parallel Sysplex Configuration, Volume 2: Cookbook

2.4.3.2 VTAM Multi-Node Persistent Session (MNPS)

Activity to the VTAM MNPS structure is high while sessions are active. State

information is updated in the structure as application data is sent or received.

This averages to three CF accesses for every MNPS application send/receive.

Outbound NLP data is kept in the structure until the session partner has

acknowledged receipt. This is done to ensure that in the event of a failure, any

unacknowledged NLPs can be re-sent after recovery. Any particular NLP is only

deleted from the structure after a session partner has acknowledged receipt of

the NLP.

There is an additional peak of activity at the beginning of any MNPS recovery

action. This occurs as all the data belonging to an application is read in. For

more information on Multi-Node Persistent Session recovery processing, refer to

the

Communications Server for OS/390 SNA Network Implementation Guide

,

SC31-8565. It is possible to have more than one MNPS structure. This would

normally only be required if the storage requirements for the structure exceed

the amount of space you can have in a single structure—that is, 2 GB. In this

case, the total MNPS storage should be divided between the structures, and one

structure placed in each CF.

Like the GR structure, the MNPS structure is rebuilt using information from each

of the connected VTAMs. So if there is a double failure of VTAM and the

structure, recovery of the structure will be impacted. Also, if any of the

applications in the failed OS/390 image were using MNPS, it would not be

possible to recover those sessions. Therefore, it is highly recommended to also

place the MNPS structure in a failure-independent CF.

MNPS Structure Placement Recommendation

Because the VTAM MNPS structure has the potential to be highly active, it is

advisable, wherever possible, to place the structure in a CF that does not

contain other high-use structures.

Also, to ensure that MNPS sessions can always be recovered following an

OS/390 failure, the MNPS structure should be placed in a failure-independent

CF.

2.4.4 VTAM Structures Rebuild Considerations

The VTAM GR and MNPS list structures support rebuild into an alternative CF,

either through an operator command or automatically when a connector (VTAM

node) detects a problem either with connectivity (which includes POR and

deactivations of CF LPs) to the structure or the structure itself.

If SFM is active, the structures will be rebuilt following

any

loss of connectivity. If

SFM is active, the structure may or may not be rebuilt, depending on the value of

the REBUILDPERCENT specified for the structure.

If there is a partial loss of connectivity to the VTAM GR structure, the VTAMs that

still have connectivity will allocate a new structure and copy the contents from

the old instance to the new one. If the structure itself is lost or damaged, a new

structure will be allocated and populated using in-storage information from each

of the connected VTAMs.

Chapter 2. A Structured View of CF Structures

87