|Shared Appendix 1: Overview|
What is IT Service Vision?
Performance evaluation is expanding to include areas such as troubleshooting problems, planning for future growth, web servers, applications support, phone systems, and help desk. IT Service Vision is a performance evaluation tool that allows you to analyze these and other technology resources more effectively by enabling you to access, manage, analyze, and present large quantities of performance data and to analyze and report on the data accurately, quickly, and flexibly.
At the core of the IT Service Vision application is the capability to read and process data from almost any data source. This is because IT Service Vision not only supports many popular data collection applications on MVS, UNIX, and Microsoft Windows NT, but it also features generic collector capability that allows you to read virtually any other data source.
IT Service Vision reads metrics in the format logged by the data collectors, processes the metrics into a performance data warehouse, calculates statistics, and generates reports. IT Service Vision can minimize the volume of data that you keep, by summarizing detailed performance data into smaller groups. You select the statistics to be calculated at each reduction level. After the statistics are calculated and the data reduced into the reduction levels, you can analyze detailed information over different time periods by running reports on a specific level of your data.
You can also customize many aspects of the application. You can use or modify table and report definitions that are supplied with this solution or you can create new table and report definitions.
Uses of IT Service Vision
IT Service Vision can be used in a variety of systems management situations within your organization. In each of the following areas, IT Service Vision can provide you with the information you need to make effective decisions about your service delivery systems.
- Performance Management (Tuning)
IT Service Vision can be used to establish baselines for system usage, identify reasons for slow response time or other indications of unsatisfactory service, and balance activity across networks and servers.
- Capacity Planning
IT Service Vision can assist you in capacity planning by allowing you to track performance and workload volume against baseline values and therefore anticipate the need for resource upgrades. You can also identify where the upgrades are needed so that you implement the most efficient and cost-effective solutions. Do you need to upgrade servers or workstations? Do they need more memory, network components, or disks? How much and where? Answers to these questions allow you to put your budget to work where it is needed most.
- Problem Determination
Using IT Service Vision, you can track network activity and processes in order to identify why your network crashed or why a process failed. You can analyze web logs to see who accesses your external web and which pages they access the most. You can also run exception reports to see which web links are returning failure messages. IT Service Vision can help you analyze your phone systems to determine the optimum number of trunk groups so that you spend only the necessary money and resources while maintaining excellent customer relations.
These are just a few of the areas where IT Service Vision can help you make better business decisions concerning your technology resources.
Specific features of the IT Service Vision application include:
The Windows NT Performance Data Log Service example in the IT Service Vision Showroom contains an example of doing this on Windows NT. For easy automation, batch jobs are implemented using an extensive library of SAS macros.
The Performance Database (PDB)
A PDB is a data warehouse that contains the performance data used by IT Service Vision. Your site may have one or more PDBs, depending on how you want to collect and store your data. You can use one PDB for each data source (collector) or you can combine data from multiple data sources in the same PDB. If you want to investigate specific performance problems or use a specific PDB for testing, you may also want to create specific PDBs for those tasks.
A performance data warehouse (or database) is a group of nine SAS data libraries and files that are managed as a single unit. These libraries include data libraries, tables, and variables that define which statistics are calculated for the data, as well as PDB options that are stored in the data dictionary for the PDB (the DICTLIB library). You must have the appropriate access permissions to create or modify a PDB; therefore, these tasks are usually performed by an administrator at your site. The administrator is usually the person who installs and manages this application, the IT Service Vision server, and your PDBs.
When the PDB is set up, the administrator that creates the PDB defines the tables, variables, and options that are used when your data is processed into the PDB and reduced into summary libraries. The IT Service Vision administrator is also responsible for managing future modifications to the PDB and making backup copies of the PDB.
When IT Service Vision is installed, the master data dictionary (PGMLIB library) contains supplied table and variable definitions. When you create a PDB at your site, you can copy the supplied definitions and use them in your PDB, you can modify the supplied definitions and save a copy to use with your PDB, or you can create entirely new definitions for your new PDB. For example, if you add/copy a supplied table or variable definition to your PDB, you may want to change defaults such as the age limits of data in each data level, the settings that specify which data are kept at detail level and which statistics are calculated at each reduction level. You may also want to create or add new variables to calculate specific statistics or to drop variables that you do not want to use.
Components of a PDB
Tables, Variables, and Definitions
The data in your PDB are stored in tables . A table is a group of stored data, statistics, and definitions in a PDB that are related to a specific performance area, such as a specific type of network activity. The table definition is a description of how to create the table and it is also stored in the PDB. The definition includes settings for the table, such as the type (interval or event), BY and CLASS variable lists, and variable definitions . The variable definitions include the variable type (numeric or character), format, interpretation labels, and the statistics that are calculated at each reduction level in the table. The table and the variables within the table are created when you run the table definition.
You can also create formula variables , which are variables whose values are calculated from the data that you collect. These variables are calculated dynamically when you access the data; the values of these variables are not stored in the PDB.
Within a table, data are stored in five libraries (or levels): detail, day, week, month, year. Each of these levels contains a view , which is used to display data from one or more SAS data sets. When you report on your data, the reports are based on the views at each level of a table in a PDB.
The data in the detail level are very similar to the data as they were recorded by your collector. The other four levels, also known as reduction or summary levels, are libraries that contain your summarized (reduced) data. Therefore, if you create a table for TCP protocol statistics, it would contain all five data levels (detail, day, week, month, year) of the TCP protocol data.
Levels in the PDB
Within the PDB, your data is stored in tables and each table contains five data libraries known as levels . These libraries are further described in Data Libraries in the PDB. When you first process your data into the PDB, it is stored in the detail level library. The data in the detail level are very similar to the data as they were recorded by your collector, with only simple transformations, such as converting continuously ascending counters into rates per second or adjusting a counter that has exceeded its maximum and restarted at zero. After the data has been processed into the PDB, you then summarize or reduce the data into the other levels of the PDB. These other four libraries are called reduction or summary levels. Each level is a separate library that contains all summarized data for a specific unit of time.
You can reduce data into the reduction levels based on the default statistics provided with a new PDB or you can customize the list of statistics that should be calculated and kept for your data at each reduction level. When the data are reduced, the specified statistics are calculated and these summary statistics (not each observation from your data) are kept at each reduction level.
You can also select the reduction levels (other than detail level) in which to keep data. You can then set a time limit on how long to keep data at each level. When the data reach a specified age limit they are aged out of the PDB.
Setting Age Limits on Data in the PDB
The detail level library can easily become very large because this library basically contains the data logged by your collection software. To reduce the space required for this library and the other libraries, your site can set age limits on the data at each level. Once data are reduced from the detail level into the reduction levels you may not need to keep the detail level data, or you may want to keep only a few days of data at that level.
When you display the data from one level in a table it is referred to as a view . A view is named using a combination of the level and the table in the form of level.table . For example, the view for the detail level in a table named HN2IFT would be DETAIL.HN2IFT. The view for the day level of the same table would be DAY.HN2IFT.
A view can also be an external detail view, if your site is directly reducing that into the reduction levels from external data sets or if your site is processing data from staged external data sets. These views are used by the PDB administrator at your site.
Data Libraries in the PDB
Each PDB contains the following data libraries. The detail, day, week, month, and year libraries are available for each table in a PDB.
The DETAIL Library
The DETAIL library contains, per table, one SAS data set and one SAS data view. Typically, a table's data set contains a group of related metrics from a specific collector. Each observation in a data set represents an event or an interval. Generally, there is a one-to-one correspondence between the observations in the detail-level data set and the data records in the raw data file. For detail level, a table's view is a view of the one SAS data set.
The DAY Library
The DAY library contains, per table, one SAS data set and one SAS data view. A table's data set contains daily statistics summarized by grouping (class) variables, such as HOUR and SHIFT. These statistics are calculated from the table's data in detail level. For day level, a table's view is a view of the one SAS data set. For example, each observation in a data set in this library may contain the statistics for that group of metrics on a given machine for a given hour of the day for a given day.
The WEEK Library
The WEEK library contains, per table, a pair of SAS data sets and one SAS data view. One data set contains statistics for the current partial week. The other contains statistics for previous weeks. A table's data sets contain weekly statistics summarized by grouping (class) variables. The statistics are calculated from the table's data in detail level. For week level, a table's view combines the pair of SAS data sets. For example, each observation in a data set in this library may contain the statistics for a given machine for a given hour of the day for a given week.
The MONTH Library
The MONTH library contains, per table, a pair of SAS data sets and one SAS data view. One data set contains statistics for the current partial month. The other contains statistics for previous months. A table's data sets contain monthly statistics summarized by grouping (class) variables. The statistics are calculated from the table's data in detail level. For month level, a table's view combines the pair of SAS data sets. For example, each observation in a data set in this library may contain the statistics for a given machine for a given hour of the day for a given month
The YEAR Library
The YEAR library contains, per table, a pair of SAS data sets and one SAS data view. One data set contains statistics for the current partial year. The other contains statistics for previous years. A table's data sets contain yearly statistics summarized by grouping (class) variables. The statistics are calculated from the table's data in detail level. For year level, a table's view combines the pair of SAS data sets. For example, each observation in a data set in this library may contain the statistics for a given machine for a given hour of the day for a given year.
The DICTLIB Library
The DICTLIB library is the PDB's data dictionary and contains: table, variable, and formula variable definitions; Kept/Not Kept status for the tables and variables in the detail level; Selected/Not Selected status for the statistics in the day, week, month, and year levels; age limits for each of the levels; and other information necessary to manage the data in the PDB. These definitions control the information in the SAS data sets in the DETAIL, DAY, WEEK, MONTH, and YEAR libraries.
The PDBWORK Library
The PDBWORK library is used by the reduction step for intermediate data. The PDBWORK library (which is permanent) is used instead of the WORK library (which is temporary) so that reduction can be restarted from a checkpoint if necessary. Sometimes PDBWORK contains temporary SAS data sets, and sometimes it is empty. If data sets are present, they are used for calculating statistics for the day, week, month, and year levels.
The COLLECT Library
The COLLECT library is used for temporary data storage as your data is processed into the PDB. For example, you can use the COLLECT library to store any staging data sets that you create for use by the generic collector software. If you select one or more of the IT Service Vision XJOBS, XPRINT, XSTEPS, XSPNJOB, XSMFINT, or XNJPURG tables and choose a spin count greater than zero, this library contains the MXG SPIN library.
The ADMIN Library
The ADMIN library can be made available to all users, and therefore it may be a convenient place to store graph and text reports, logs, and so on. If you are using the data filtering routines to filter duplicate data when you process your data into the PDB, this library is used by those routines. (For more information on data filtering, refer to the Processing Duplicate Data topic in the Macro Reference or see your IT Service Vision server administrator.)
IT Service Vision Software Components
Before you collect, analyze, or report on data using IT Service Vision, you should become familiar with the structure of the software. As you access and view the data in your PDB, you will be using the following libraries and software components. If an item is platform-specific, the platform is indicated below.
When you gather data at your site, you will do so using a data collection application which is referred to as the collector. This collector could be an application that is supported by IT Service Vision, or it may be any other software that you are using to gather data.
When data are copied into the PDB from your raw data file, this task is referred to as processing your data or the process task. When data are copied into the PDB, the process task sorts and organizes the data into the detail level. When you then summarize the data into the non-detail levels in the PDB (day, week, month, and year) this task is referred to as reducing the data or the reduce task.
Typically a site has one PDB for each collector that is in use, but this is not required. You can use a single PDB for multiple collectors or you may want to create a separate PDB for investigating specific performance problems or for testing.
The Data Dictionary (DICTLIB Library)
Each PDB contains a data dictionary , which is a SAS library within the PDB, known as DICTLIB. The PDB's data dictionary contains general PDB options as well as your specifications about which data to keep in the PDB. For example, your data collection software may collect some data that you do not want to keep. In this case you can specify, within the data dictionary, for the data to be dropped.
The Master Data Dictionary (PGMLIB Library)
PGMLIB, which contains the master data dictionary, is the SAS data library in which most of the IT Service Vision software for your current host resides. It contains: the master data dictionary (supplied table and variable definitions), supplied report definitions, online help, and the IT Service Vision client and server executables.
Note: On some hosts, additional software for the user interface is installed in the CPSYSLIB library. If you have more than one release level of IT Service Vision installed, then you will have a separate version of PGMLIB for each version of the software.
The master data dictionary contains a table definition for each supported type (topic) of data. A table definition is used to describe the table to IT Service Vision, including the type of data it contains, the data source, and so on. For each field in the record, the data dictionary also contains a variable definition that describes the field, its data type, and so on.
The PGMLIB library is created when you install IT Service Vision. On UNIX it is installed in the directory !sasroot/saspgm/cpe/pgmlib . On MVS the location is sas.cpe.pgmlib . On Windows '95, Windows NT, and OS/2, the location is !sasroot\cpe\pgmlib .
The terms !sasroot or sas represent the location or prefix where SAS is installed on your current host. When IT Service Vision is updated, the libraries and information in PGMLIB are updated. You must have write access privileges to add or modify any of the items in PGMLIB. Therefore, PGMLIB updates or modifications are usually performed by the IT Service Vision administrator at your site.
When you create a new PDB, it contains supplied table definitions, supplied report definitions, and a supplied sample data dictionary. The IT Service Vision administrator at your site can customize those existing definitions, create new definitions, and/or add table and variable definitions from the master data dictionary (PGMLIB). By installing user-generated table definitions in the master data dictionary, your IT Service Vision administrator can easily add those definitions to other PDBs at your site.
You do not have to have write access to PGMLIB in order to view or search the list of supplied table definitions, variables and formula variable definitions, and report definitions.
For more information about modifying supplied table and variable definitions, refer to the IT Service Vision Setup documentation.
The SITELIB library is a SAS data library that contains your site-specific information, such as site holidays, work shifts, and formats. For example, supplied formats and site formats are stored in SITELIB.CPFMTS. By saving site-specific information in one central library, all users at your site can access these formats and settings.
A default SITELIB library is created when you install IT Service Vision. On UNIX it is installed in the directory !sasroot/saspgm/cpe/sitelib . On MVS the location is sas.cpe.sitelib . For Windows '95, Windows NT, and OS/2, the location is !sasroot\cpe\sitelib . The terms !sasroot or sas represent the location or prefix where SAS is installed on your current host.
If you are processing data for multiple sites or groups within your organization, then you may have a separate SITELIB library for each site or group.
If your organization wants to change the defaults in SITELIB to customize one or more options, you should copy the default SITELIB library and make changes to the copy. To modify settings or add items to SITELIB, you must have write access permission for this library. These modifications are usually performed by an IT Service Vision administrator or someone else with write permission to this library. For more information on accessing SITELIB or setting site specific preferences such as holidays and shifts, refer to the IT Service Vision Setup documentation.
The SASUSER library is a SAS data library that is used by your interactive SAS session for storing user-specific information, such as the locations of PDBs that you have used and any customizations that you specify for SAS options.
When you start the SAS system, a SASUSER library is automatically created for you, if one does not already exist. You do not need to install or maintain this library. When you customize keys, menus or other items within the application, these modifications are automatically saved in your SASUSER. By default, any source, log or report files that you save are saved in this library.
If you delete this library, it is created the next time that you start SAS or IT Service Vision. However, you will need to reset any customizations that you want to use.
Note: By default, this library is not used in batch on MVS. For platform-specific information on the SASUSER library, refer to the SAS Companion documentation for your platform and your current version of SAS.
The MISC directory on UNIX and the SASMISC directory on Windows 95, Windows NT, or OS/2, contains the following information: examples for running IT Service Vision, the autoexec file, and the cpe command for starting SAS and this application. By default, this library is located in the directory where SAS is installed on the current platform. For example, the default location for UNIX is !sasroot/misc/cpe and the default location for Windows NT, Windows '95, or OS/2 is !sasroot\cpe\sasmisc.
The term !sasroot represents the location where SAS is installed. If you need to modify any of the files in the MISC or SASMISC directories, you should copy the file from this directory and modify the copy.
The CPMISC directory is described later in this section.
IT Service Vision Macros
IT Service Vision provides macros that can be run in batch. Macros that require write access to the PDB, such as %CSPROCES, %CMPROCES, %CPREDUCE, or any macros that update the data dictionary, must be run on the IT Service Vision server. Reporting macros can be run from the IT Service Vision Server.
The IT Service Vision interface is built based on these macros so that when you select tasks within the application interface, the macros execute to perform the tasks. However, if you want to run jobs in batch mode you can also write batch jobs that include calls to these macros. On UNIX the macros are located in the SAS autocall library in !sasroot/sasautos . On MVS the macros are located in sas.autolib . On Windows '95, Windows NT, and OS/2, the macros are located in the directory !sasroot\cpe\sasmacro .
The terms !sasroot or sas represent the location or prefix where SAS is installed.
The IT Service Vision macros are documented in the Macro Reference, which is available by selecting the following menu path:
OnlineHelp -> Documentation Contents -> Usage and Reference Macro Reference
SCHEMA and MIBS Directories
On a UNIX or PC (Windows 95, Windows NT, or OS/2s) platform, these are the directories that contain the schema files and MIBs, used by various network and systems management platforms. On UNIX these libraries are in directories !sasroot/saspgm/cpe/schema.tar and !sasroot/saspgm/cpe/mibs.tar . On the PC these libraries are located in !sasroot\cpe\schema.zip, !sasroot\cpe\schema.zip, and !sasroot\cpe\mibs.zip .
The term !sasroot represents the location or prefix where SAS is installed.
For IT Service Vision MVS, MXG software is packaged with IT Service Vision. Your site can customize the MXG code and also utilize the SAS formats provided by MXG, so that the data values are formatted as they are read in and written to the data sets. MXG software consists of three libraries:
This library contains the SAS macro code used to build the staging SAS data set for each table. This is a standard MVS PDS (partitioned dataset) and is created when MXG is installed.
This library contains any modified versions (for example, to make use of exits) of the members in MXG.MXG.SOURCLIB. This is a standard MVS PDS, just like MXG.MXG.SOURCLIB, but is created by your site's IT Service Vision/MXG administrator.
This library contains the SAS formats and informats that are used by MXG. When MXG is installed, this SAS library is created by running the JCLINSTL job in MXG.MXG.SOURCLIB.
Note: MXG uses the ddname SOURCLIB for the concatenation of SOURCLIBs. The order of the concatenation is important. For any given member, MXG uses the first member that it encounters in the concatenation. Thus, the intended order is MXG.USERID.SOURCLIB and then MXG.MXG.SOURCLIB. MXG uses the ddname LIBRARY for the SAS library named MXG.MXG.FORMATS.
If you are using IT Service Vision macros on MVS, you may need to know the locations of these MXG libraries. MXG setup information is provided in the IT Service Vision Setup documentation.
The CPMISC data set is a standard MVS partitioned data set (PDS). By default, it is installed in sas.cpe.cpmisc where sas represents the prefix where SAS is installed at your site. CPMISC contains information such as:
JCL examples for processing, reducing, and reporting on supported data collectors in batch.
CMAUTOEX, which is the SAS autoexec that contains startup parameters for IT Service Vision.
CMCPE clist, which is the default name for the clist with which you start SAS and automatically invoke the IT Service Vision server on MVS.
A list of all of the members in this PDS and a short description of each member's content is in the $README member.
If you or the IT Service Vision administrator at your site needs to modify a member from CPMISC, copy the member to another location and modify the copy.
UNIX and PC platforms use the MISC directory described earlier in this section.
The Data Collection and Analysis Process
The routine tasks that you will perform to process data and view reports can be summarized as follows:
- Collect data that contain your performance information. The data files are collected from the software that you are using to gather your raw data. Your collector, whether supported by name or by means of the generic collector capability, must be set up to gather data before you begin using this application.
- Process raw data into the detail level tables of your performance data warehouse (PDB). Your site can set up a daily job to process the data. When detail level data reach a specified age (the amount of days, weeks, months, etc. that you specify), they are automatically discarded or aged out.
- Reduce the detail level data into one or more reduction levels in the PDB: daily, weekly, monthly, or yearly. In most cases, you or an administrator at your site will set up a daily job to reduce the data immediately after processing log files. You can specify how long you want to keep the data. When non-detail-level data reach the specified age, they are automatically discarded.
- Run the supplied report definitions or create and run your own custom report definitions to help you analyze your data and identify performance issues. You can generate reports on any level of data in the PDB, from detail level to any of the summary levels.
Before you begin collecting and processing data into the IT Service Vision PDB, an administrator at your site must perform the setup procedures provided in the IT Service Vision Setup documentation.