Flow Map Manual

3 Flow Map Host
098935 Tandem Computers Incorporated 3–1
This section discusses the Flow Map Host portion of the Flow Map product. It
includes information on:
Understanding the Flow Map Host components
Selecting objects
Creating a configuration file
Customizing the CONFIG file
Creating Flow Map reports
Understanding the
Flow Map Host
Components
Flow Map Host (FMH) is the component of the Flow Map product you must run first.
It takes a consolidated data file from Tandem Performance Data Collector (TPDC) and
builds a comma separated values (CSV) file to hold the resultant objects and
connection information. Finally, the CSV file becomes the input for FMP.
Objects Flow Map Host generates Processing and Database objects by combining performance
counters extracted from the Measure entities. Each object class has three object types.
PROCESS Group, Program, and PROCESS are the object types for Processing objects.
FILE Group, FILE, and DISKFILE are the object types for Database objects.
Processing Object A Processing object gives a complete view of a processing environment, including a
summary of all the input-output (I/O) performed by the processes concerned.
Object Types
A PROCESS Group consists of PROCESS object counters grouped by your own
criteria. Typical criteria include Program file names, PIDs, User IDs, and Creator IDs.
A PROCESS Group is the sum of all the counters from all the PROCESS objects that
you selected for the group.
A Program is a group of PROCESSes with the same Program file name. It is the sum
of all the counters from PROCESS objects run from the same Program file.
A PROCESS object is based on a single Measure PROCESS entity. All of the counters
of all entities associated with the PROCESS are added to make a single PROCESS
object. Typical entities are: FILE (type 0), FILE (type 3), FILE (other), and DISCOPEN.
For example, if the PROCESS opens three logical disk files, then the counters from the
three FILE (type 3) TPDC entities are added to form that part of the Flow Map
PROCESS object.
FILE counters are split into three subsets: Disk data, process-to-process data, and
other file IO. These sets relate to type 3 entries (disk), type 0 (process), and all other
types of FILE entities.