|
|
This chapter provides an overview of the statistics collection process implemented by the Cisco StrataView Plus (SV+) Statistics Collection Manager (SCM). This application, which is started from the Cisco SV+ Main Menu, serves two purposes:
The Cisco SV+ SCM should be left running at all times when you want collection to occur continuously. Unlike other Cisco SV+ features, statistics collection is managed from within the Cisco SV+ SCM program, not from the Cisco SV+ core processes. When you exit the Cisco SV+ SCM program, collection does not occur until you restart the Cisco SV+ SCM.
Note, you can distribute the statistics collection workload among multiple workstations. The Cisco SV+ SCM CD-ROM you received with your Cisco SV+ 9.1 installation package, contains a stand-alone version of the Cisco SV+ SCM, and can be installed on one or more workstations, without the Cisco SV+ core processes or GUI applications. This stand-alone Cisco SV+ SCM program can be used to collect statistics from some or all nodes in your network, thereby off loading statistics collection from the Cisco SV+ workstation.
The Cisco SV+ SCM is not the application to use for viewing statistical information, however. That functionality is provided through the Reports and New Reports applications. The Reports application uses the Wingz application to display the data gathered by the Cisco SV+ SCM. The New Reports application displays network resources and performance statistical data in Cisco SV+ generated windows.
For information about the Wingz-based Reports application, see Chapter 5 "Reports Application".
For information about the New Reports application, see the Chapter 10 "New Reports Application".
The Statistics Manager continually polls nodes for a datafile which is then moved to a the /usr/users/svplus/incoming directory.
A parser daemon in the Cisco SV+ core processes periodically polls this directory. When it finds new information at that location, the file is parsed and the data written to the Cisco SV+ Informix OnLine database. This parser uses the file parserconfig.stats that is generated by the Network Parameters window shown in Figure 4-3. You should set these network parameters from the workstation on which the core processes are installed.When you set them from one of the stand-alone versions of the Cisco SV+ SCM, you must copy the file onto the machine on which the Cisco SV+ core process are installed. This file is located in the usr/users/svplus/statsmgrdir directory on both machines.
Nodes must be selected, and specific statistics enabled, before the SCM can collect data. You configure nodes and statistics using the options on the Statistics Manager window's Config menu. (See "The Statistics Manager Window" described later in this chapter.)
When you configure statistics for collection, the Statistics Manager invokes the Trivial File Transfer Protocol (TFTP) put command to each enabled node.
Each node, when enabled for statistics, collects data according to individual configurations for collection interval, duration of retention, and bucket size. For time synchronization, a telnet command is sent to the time sync node connected to the Cisco SV+ workstation, to obtain the current GMT (Greenwich Mean Time) of the node, which is then used for the statistics time stamp.The Statistics Manager contains a built-in watchdog process, which automatically restarts the application in the case of an abnormal termination.
Preparation for statistics collection includes:
1. Review and possible modification of parameters for the statistics collection process.
2. Determination of the types of statistics to be collected from each node.
3. Distribution of nodes among multiple installations of the Cisco SV+ SCM, when stand-alone versions are to be used.
The Status Display pane in the Statistics Manager window (shown in Figure 4-2) lists some, but not all, of the possible parameters you may want to consider setting. To view all of the possible parameters, and to change any of them, select the Network Parameters option from the Config menu. The Statistics Parameters window is displayed. Most of the parameters can use the default values. The parameters you should review are
Time Synchronization Node | Select the node supplying the master system timing. When possible, it should be set to the attached gateway routing node where SCM is installed. When your network crosses over time zones, it is important that all nodes in all time zones remain synchronized to a single clock (namely, the clock in the node defined as the Time Synchronization Node in the Statistics Manager's main window). |
Peak Statistics Enable | Toggle to OFF unless peak statistics are required. |
Statistics Collection Period | Collection interval display. |
Bucket Interval | This must not exceed the value set for the collection period. |
Purge Old Files | Recommended unless doing a long-term study on only one, or a few, statistics. |
Before statistics collection is enabled, use the cnfdate or cnftime commands to configure all the nodes in your network to the same time as the Time Synchronization Node. Use the cnftmzn GMT command to set the node time to GMT (Greenwich Mean Time). Once all nodes are using the same time as the Time Synchronization Node, you can use the cnftmzn command to set the node time to whichever time zone you want, however, the time must come from the Time Synchronization Node.
File directories are assigned a default name by the system. These fields may be renamed to something more appropriate to what is being collected. These parameters should be configured before any statistics collection is initiated. They can be changed later, but the statistics collection process must first be stopped.
Prior to installing Cisco SV+, you should decide the types and quantity of statistics to be collected. Configurations for statistics in larger networks may involve trade-offs to accommodate procurement of high-volume statistics, as well as to maintain sufficient node memory allocation, Cisco SV+ link bandwidth, or Informix database disk capacity for statistics.
When a node is added to a network, some statistics are enabled by default. You can disable any or all of these, when you don't want them collected. However, when you delete a node and then add it back into the network, the default set of statistics are once again enabled (provided no stats are enabled in this node).
For each statistic, you can select the bucket interval (also called the bucket type and sometimes granularity of the sample), the collection interval, and retention interval.
Bucket interval | Specifies the interval over which the node accumulates a sample (that is, every 5, 10, 15, 30, or 60 minutes). When all buckets are full, the collection wraps around and fills the first bucket. The number of buckets retained is constrained by the amount of memory allocated for this process. Default: 60 minutes. |
Collection interval | Specifies the frequency with which Cisco SV+ collects these samples. The collection interval should always be set equal to, or longer than, the bucket interval. Default: 60 minutes. |
Retention interval | Specifies how long the data is maintained by the Cisco SV+ database. The default retention period is 24 hours. The maximum retention period is 8960 hours (1 year). |
The Statistics Manager window is displayed from the Cisco SV+ Main Menu by specifying 6 (Start Stats Mgr GUI) and pressing Return.
The StrataView Statistics Manager window is displayed.
The Status Display pane lists some, though not all, of the parameters that can be set to govern the network collection process.
The Outgoing Requests pane, Incoming Responses pane, and Errors pane display messages recording the collection process.
You select this option to close the Statistics Manager window. Selecting this option displays a message reminding you that exiting the Statistics Manager halts the entire collection process. When you click on the OK button, the Statistics Manager is shut down and statistics collection stops.
You select this option to display the Network Parameters window, where you can change the parameters that govern the collection process (including such things as maximum number of concurrent requests, statistics collection intervals, and target directories for incoming statistics). See "Network Parameters Window" for a detailed description of this window's components.
You select this option to display the Statistics Node Selection window, which is where you start or stop the collection process for individual nodes,. See "Node Selection Window" for a detailed description of this window's components.
You select this option to display a submenu with these additional options:
You select this option to clear the Errors pane in the Cisco SV+ Statistics Manager window.
The Network Parameters window is displayed when you select the Config menu's Network Parameters option in the Statistics Manager window.
This setting controls the number of child processes created. Each child process is responsible for one file transfer at a time, which indicates the number of concurrent requests to the network. The default is 3.
You specify the number of requests allowed per subnet. This setting is used to throttle the number of requests being retrieved from the network on a particular access point. Statistics enable files are not considered part of this throttling. The default is 3.
This field is used with stand-alone installations only. You specify the duration for child process to sleep prior to each request. The default is 0.
This field is used with stand-alone installations only. You specify the maximum packets to be transmitted from the child process. A packet is fixed at 512 bytes in TFTP specifications. A value of 500 transfers packets at up to 256K/second. The default is 0.
An entry of 0 (zero) indicates unlimited transfer rate.
You specify the number of retries the child process makes in attempting to get files from the network. The default is 0.
You specify the time-out value (in seconds) for TFTP GET requests. Increase this value on busy networks. Busy networks should be configured for a high value. The default is 60 seconds.
You specify the name of the gateway routing node, which the Cisco SV+ SCM uses for time synchronization. All nodes in all domains must be set to GMT using the cnftmzn command.
You specify the interval value (in minutes) in which statistics are gathered from the network. The default is 60 minutes.
You specify the Bucket Interval Period value (in minutes) to be kept for the single bucket on the node.
Toggle this button to set peak values on network statistics to either ON (default) or OFF.
You specify the total number of history files to collect. When, for example, you select 5, and the interval is set at 60 minutes, the current hour, and the prior 4 hours are tracked for collection from the network. This "history" on collection completion or failure is tracked for 5 files (1 file = 1 collection interval). It is recommended you select at least 5 for this setting. The default is 5.
You specify the number of days to keep old statistics. Statistics older than the current date/time, less this field in days, are automatically deleted from the database. The default is 30 days.
A value of 0 disables statistics deletion.
You specify the filename to use as temporary storage for incoming files.
You specify the name of the incoming file directory used to store incoming files to be parsed. The incoming and temporary directories must be within either the same UNIX partition or NFS mounted partition.
You specify the value (in days) used to determine which old files are purged from the /usr/users/svplus/spool directory when space in your hard disk's Incoming partition drops below 20 megabytes. The files with a date `N' days older than today's date are purged. (0 = no purging is done.) The default is 2 days.
This toggle controls the ability to purge files as they are parsed from the statistics parser. You set it to OFF to move files to the /usr/users/svplus/spool/save directory. Currently, the save directory is not configurable. We recommend you delete old files. The default is ON.
The Node Selection window, shown in Figure 4-3, is displayed when you select the Config menu's Node Selection option in the Statistics Manager window.
The node filtering feature, used only by a stand-alone installation of the Cisco SV+ SCM, is used to reduce the nodes listed in the Unselected Nodes pane to those which meet user-defined filter conditions. Enter a string into the Add Filter Items field, then click on the Add Filter button, or select a filter item from the list in the larger pane and click on the Delete Filter button. Click on the Rescan Node File button to apply newly defined filters.
Displays all nodes for which statistics can be collected. Select one or more nodes and click on the Selected button in the column to the right of the pane, to add them to the Selected Nodes list. Or, use the All button to add all nodes to the Selected Nodes list.
Displays all nodes Cisco SV+ needs to collect statistics from. To add nodes to this list, select the node from the UnSelected Nodes list, then press the Selected Nodes right-arrow button or press the ALL nodes right arrow button. To remove a node, select in from the list and press the left-arrow select button, or use the All button to clear the Selected Nodes list.
The "@" symbol at the end of a node name implies SCM is collecting statistics from this node.
Determines what action to apply to the nodes in the Selected Nodes list. Click on one of the following buttons:
Applies the selected action to the nodes in the Selected Nodes list. The Statistics Node Selection window remains displayed.
Applies the selected action to the nodes in the Selected Nodes list, then closes the Statistics Node Selection window.
Closes the Statistics Node Selected window without taking any action.
The Enable/Disable window, shown in Figure 4-6, is displayed when you select any of the Config menu's Stats Enable option in the Statistics Manager window: Connections, Service Interfaces, Trunks, and Ports.
The window operates the same for each of the four menu options, except data in the window's fields varies according to the menu option selected.
You click on the small button on the right to display a list of the object subtypes for the Stats Enable Object Type selected.
The peak values set on all statistics within a subtype are displayed in this pane. It is applicable only when you enable the Peak Values toggle button in the Network Parameters pane. You modify this value by specifying a new value in this field.
A list of all unselected statistics types available for this subobject type is displayed in this pane.
A list of the selected statistics types for this subobject type is displayed in this pane. To select a statistic, click on the desired statistic in the UnSelected Statistics pane, then click on the Selected Stats right-arrow button. To select all stats, click on the ALL Stats right-arrow button. To remove a particular statistic, click on the desired statistic in the Selected Statistics pane, then click on the Unselected Stats left-arrow button. To disable all stats for the current object, click on the ALL Stats left-arrow button.
To view the default list of statistics enabled or disabled for collection and to modify this list, select the Stats Enable option from the Config menu of the Statistics Manager window, to select connections, service interfaces, trunks, or ports.
| Connections | Service Interfaces | Trunks | Ports |
|---|---|---|---|
Voice | T1 | Narrowband | Frame Relay ASI Port |
Data | E1 / J1 | IPX ATM | FastPAD ASI Port |
Frame Relay | ASI line | BPX ATM | AXIS Frame Relay Port |
FastPAD Voice | AXIS E1/ T1 | AXIS Narrowband | ATM |
FastPAD Switched Voice | AXIS E1/ T1 | AXIS Narrowband | ATM |
FastPAD Data | AXIS E1/ T1 | AXIS Narrowband | ATM |
FastPAD Frame Relay | AXIS E1/ T1 | AXIS Narrowband | ATM |
Use the Selected Stats or ALL Stats buttons in the Statistics Configuration window to move the statistics to be enabled to the Selected Statistics list on the right side of the window.
Use the left-arrow buttons to remove statistics you no longer want enabled from the Selected Statistics list. Statistics lists can be created without interrupting normal statistics collection from the network.
You must prepare a list of all nodes in the network that will be collecting statistics for display on this Cisco SV+ workstation. When there is only one Cisco SV+ workstation attached to the network, this list consists of all nodes in the network. Every node without a LAN IP address must have an IP Relay IP address assigned. Use the cnfnwip command to assign IP Relay addresses.
Use the Telnet protocol to configure nodes, using the following commands:
Use this command to configure an IP address and subnet mask for the node. Network statistics are collected by the Statistics Master process resident in Cisco SV+. The Statistics Manager requests/receives statistics using TFTP Get and Put messages. These TFTP messages are transferred between the node and the Statistics Master via IP Relay. This command defines the IP address used to route the messages from/to the node.
Use the following for IPX, BPX, and IGX:
Full Name | Configure network node IP address. | |
Syntax | cnfnwip parameters |
|
Related Commands | cnfstatmast, dspnwip |
|
Attributes | Privilege | 1-6 |
| Jobs | No |
| Log | Yes |
| Node | IPX, BPX, and IGX |
| Lock | Yes |
Example 1 Description | cnfnwip 199.35.96.217 255.255.255.0 Configure 199.35.96.217 as the IP address and 255.255.255.0 as the subnet mask for the node. | |
For AXIS, send a cnfifip command with no parameters to generate a list of example parameters, then resend the cnfifip command with the appropriate parameters.
Nodes whose statistics are collected by a stand-alone installation of the Cisco SV+ SCM, and for which you have added an IP address using the cnfnwip command, must also be added to the Cisco SV+ /etc/hosts file so the Cisco SV+ SCM can use TFTP.
The Cisco SV+ workstation must also have an IP address assigned. Use the cnfstatmast command to assign an IP address to the Cisco SV+ workstation and define it as the statistics collection point. Each node to be managed by the Cisco SV+ workstation must run this command in order to route messages to and from the Cisco SV+ workstation. The cnfstatmast command can be included in a job to automate the process.
This command configures an IP address for the Statistics Master process resident in Cisco SV+. Network statistics are collected by the Statistics Master process via TFTP. The Statistics Manager requests/receives statistics using TFTP Get and Put messages. These TFTP messages are transferred between the node and the Statistics Master via IP Relay. This command defines the IP address used to route the messages from/to the Statistics Master. See the cnfnwip command for details on setting a nodes address.
Full Name | Configure Statistics Master Cisco SV+ Address | |
Syntax | cnfstatmast <parameters> |
|
Related Commands | cnfnwip, dspnwip |
|
Attributes | Privilege | 1 |
| Jobs | Yes |
| Log | Yes |
| Node | IPX a |
| Lock | Yes |
Example 1 Description | cnfstatmast 199.35.96.217 Configure 199.35.96.217 as the IP address for the statistics master. The IP address is a 32-bit address specified in x.x.x.x format where each x is a value from 1 to 255. | |
For large networks, there may be multiple Cisco SV+ workstations, each one managing a part of the overall network. A separate list needs to be prepared for each Cisco SV+ workstation. IP addresses must be configured as described above.
For AXIS, use the cnfstatmgr command.
The following system constraints apply:
To model a system, follow these steps:
Determine your requirements (that is, given no constraints, the statistics you want to capture).
Step 1 Using the default parameters for bucket interval, number of buckets, collection interval and collection request rate, are the system bounds met? Calculate the IPX memory requirements and collection request rate based on the bucket interval and collection interval to see when the desired set of statistics are available without reconfiguration.
When they are, then the desired subobject statistics classes need only be enabled.
When they are not, the following parameters may need to be modified before the statistics can be enabled:
When IPX memory is the critical parameter, the bucket interval may be increased:
When the collection request rate is the critical parameter:
Step 2 Has the parameter modification allowed collection of the desired statistics?
If so, modify the selected parameters and enable the statistics.
If not, it may be necessary to distribute statistics collection over a number of Cisco SV+ workstations in the network. It should be noted, the situation in which two Cisco SV+ workstations each collect half of a node's statistics provides a better aggregate collection capability than when a single Cisco SV+ workstation collects all of one node's statistics.
You can install stand-alone SCMs on as many workstations as you like in order to distribute the workload of statistics collection.
You can also use multiple installations of Cisco SV+, up to a maximum of 12 installations per network, and use these installations to distribute the workload or to provide duplicate statistics collection as a backup.
However, nodes cannot differentiate between installations. A node gathers information for all enabled statistics and writes the data to a single location. You cannot have one installation collect some statistics and another installation collect different statistics from the same group of nodes.
Posted: Mon Oct 4 19:49:32 PDT 1999
Copyright 1989-1999©Cisco Systems Inc.