When VTAM is started, options can be used to define the initial VTAM network and to select optional VTAM facilities. These start options can be specified by the operator as part of the START command (in OS/VS only) or as a response to the prompting of VT AM. The options can also be predefined and fIled on the VT AM definition library.
(See "Starting VT AM," in Chapter 4, for information on specifying options via the system operator's console.)
The VTAM start options tailor VTAM to the installation's needs each time the telecommunication system is started. Predefining start options relieves the network operator of this activity. In addition, more than one version of the start options can be predefined, each version specifying a different VT AM configuration. With different sets of predefined options, the installation can initialize a particular VT AM system merely by selecting the appropriate set of options.
The predefined start options are stored under the member name ATCSTRxx (where xx and a data set must be filed under this name even if it does not contain any options).
Other ATCSTRxx data sets can be created, but the specific data set required is specified by the installation when VTAM is started.
The following start information can be supplied in the ATCSTRxx data set:
• Which major and minor nodes are to be traced by the VTAM trace facility
• Which major nodes are to be activated during start processing (VT AM initialization)
• The size of VT AM storage pools and in OS/VS, which pageable storage pools should be fIXed
• The maximum number of major nodes active at anyone time
• Whether certain network operator messages are to be suppressed
• Whether prompting messages should be sent to the network operator to obtain additional start information
• Whether the VT AM logon monitor facility for start-stop and BSC terminals (referred to as the network solicitor) is to be started when VT AM is started
If the network solicitor is activated, it begins monitoring the active terminals assigned to it for logons. See "The Network Solicitor," in Chapter 8, for a description of the network solicitor.
Chapter 3. Creating a Telecommunication System with vr AM 49
Defining VT AM Buffering
The trace facility is activated for specified terminals, and the trace continues for as long as the terminals are active or until the network operator stops the trace. See "Starting and Stopping VT AM Facilities," in Chapter 4, for information on stopping the VT AM traces.
See "Serviceability Aids," in Chapter 6, for a description ofVTAM's trace facility.
The names of major nodes to be activated when VT AM is started are stored as a member of the VT AM defmition library. These names must be stored under the member name ATCCONxx (where xx is a two-character identification created by the installation).
ATCCONOO is a default member name; other ATCCONxx member names must be specified by the installation during start processing. All ATCCONxx members must be created and fIled by the installation. If an ATCCONxx member is not created and fIled, the default member name ATCCONOO will be fIled and a warning message will be provided when VTAM is started. See "Major and Minor Node Structure," earlier in this chapter, for a definition ofa major node.
To activate major nodes during VTAM start processing, an ATCCONxx member is specified as a start option. This member then contains a list of the major nodes to be activated. This option enables the installation to specify the initial telecommunication configura tion.
VTAM uses storage pools to allocate space for control blocks, buffers, and channel programs. Pools are established in both fixed and pageable storage. See "Defining VT AM Buffering," later in this chapter, for more information on specifying storage pools and on VT AM's use of these pools.
If the network operator is to be prompted, VT AM transmits messages to the system operator's console requesting that start options be entered from the console. The network operator is prompted only (1) if prompting is requested in the ATCSTROO data set and (for OS/VS) no start parameters are entered with the START command or (2) if an error is encountered by VT AM during VT AM initialization. See "Starting VTAM," in Chapter 4, for information on the role of the network operator in starting VT AM.
An installation can prepare and file multiple start-parameter members and multiple node-name members. Various combinations of these parameters can thus be used when VT AM is started to initialize VT AM systems with different characteristics.
VT AM is not a queued access method, such as TCAM; VT AM does not save a message until the message is requested by a terminal or an application program. However, VTAM can hold data temporarily until a VT AM application program requests it or until it can be sent to a terminal. Figure 3-11 shows how VT AM buffers handle input and output data.
VT AM has both flXed and pageable storage pools for buffering data. In addition, storage pools are required for the VT AM control blocks that keep track of incoming data that has not yet been requested by a VT AM application program and to keep track of individual VT AM application program input and output requests. Each storage pool serves all active application programs. Because installation needs may vary, VT AM allows an installation to specify the size of each of these storage pools. In OS/VS, the installation can fix one or more of these pools in main storage. The names of each of the storage pools are different in each operating system and may be found in the appropriate VT AM System Programmer's Guide.
When an installation starts VT AM, it specifies the number of buffers (elements) i!1 each storage pool, the size of each buffer, and for the data storage pool, a threshold number of buffers. The threshold number is the point beyond which, in OS/VS, requests for buffers for input or output are queued until buffers in use become free. Where no value is
INPUT VTAM
Application Program
• •
RECEIVE
•
VTAM
--- -:r::=----
11 Data arrives from a logical unit independent of any VTAM application program request for data. It is placed in a buffer in the fixed storage pool of data buffers.
2 If there is an outstanding request for data from the logical unit, the data is moved from the buffer in the fixed storage pool to the specified VT AM application program data area (shown with dotted lines).
If there is no re"quest yet for the data, it is 11)0ved from the fixed to the pageable data buffer pool. Later, it can be paged out of main storage.
3 When an input request is issued, the data is paged into main storage, if necessary and moved to the specified VT AM application program data area.
OUTPUT VTAM
Application Program VTAM
•
Fixed Pool• ~ I
SEND
•
L
l r
1J
2'"
)v
1 The VTAM application program requests data transmission. When VTAM has sufficient buffers in the fixed storage pool, }i moves the data to the fixed pool.
2 A short time later, the data is sent to the NCP or local terminal, freeing the buffers.
Figure 3-11. How VTAM Buffers Input and Output Data
Chapter 3. Creating a Telecommunication System with VTAM 51
APPL'I VTAM
I
/ / /
APPL2
~
I
/ /
-/ / /
/
Storage Pools
/ / / / / /
/
1
I / Storage Pools for Control Blocks
I
I I I I I I
~
, /
I
TERM1
NCP
/ - - Each storage pool has a special purpose and is shared by
,
I- all active VTAM application programs. The number of buffers in each pool, the size of each buffer, and a buffer threshold are specified when starting VTAM.
, ,
, ,
,
~I---~, ,
, ,
, I
, ,
, ,
, I
, ,
I :storage Pools for Data Buffers : I
The maximum number of buffers for a particular application program to terminal connection (for example, APPL 1-LU1) in the pageable data buffer pool is specified when defining VT AM.
Figure 3-12. How an Installation Controls Control Block and Data Buffer Storage Pools in VTAM
specified, VT AM supplies a default value. In DOS/VS, VT AM supplies the threshold value; any number specified by the installation is ignored. If the threshold is consistently exceeded, program performance and terminal response time may be affected. An installation may have to "tune" the buffer 'number, size, and threshold of its storage pools to achieve an efficient combination of storage utilization and performance.
In addition to specifying the size and threshold of storage pools when starting VTAM, the installation can control the amount of buffer storage for data that can be used for each individual connection between an application program and a terminal. Controlling buffer storage for connection between application programs and terminals prevents an individual connection from monopolizing data buffers and slowing down communication between other connections. The number of data buffers for each connection is specified during VTAM defmition. VT AM determines the data buffer threshold for each connection by multiplying a number specified for the application program (in the APPL statement) by a number specified for the terminal (in a LOCAL, LV, TERMINAL, COMP, or VTERM statement). For incoming data from terminals, if the threshold is exceeded, VTAM clears any data that may have arrived from the terminal, issues a Clear command if the terminal is a logical unit, and schedules the VT AM application program's LOSTERM exit routine.
The program is resonsible for resynchronizing the connection. This situation will not occur, however, if the installation ensures that the VTAM application program requests input at a rate that is not below the rate at which data may be coming in to VT AM.
Figure 3-12 shows how an installation controls the buffering of data.
Although VT AM buffers both data coming in from a terminal to a VT AM application program and data going from a VT AM application program to a terminal, input data may have to be buffered for a longer time than output data. For input data, VT AM must buffer the data until it is requested by a VT AM application program. For output data, VT AM decides when it wants to begin handling the data; after receiving an output request, VT AM need not move the data from the VT AM application program until it has ensured that it has buffers for it. For this reason, more planning is required for the buffering of input data. In general, a VT AM application program should be written so that at least one request for input is always outstanding, or at least so that very little time elapses during which no request for input is outstanding. This may reduce the possibility of exceeding the buffer size for data and control block storage pools.
In OS/VS, an installation has the option to fix VT AM storage pools. An installation that requires high performance may want to fix those pools that may not always be resident because of infrequency of use.
Recommendations and guidelines for determining the sizes and thresholds for storage pools of buffers are provided in the VT AM System Programmer's Guide and storage estimate publication for the appropriate operating system.
Chapter 3. Creating a Telecommunication System with VTAM 53