ovowin_75_tsg

Upload: carlos-marrero

Post on 06-Apr-2018

212 views

Category:

Documents


0 download

TRANSCRIPT

  • 8/3/2019 ovowin_75_tsg

    1/91

    HP Internal Only

    Version 4.0

    Edition Date: 05/ 06/ 2005

    HP OpenView

    Windows Solution

    TechnicalSales

    Guide

  • 8/3/2019 ovowin_75_tsg

    2/91

    Copyright 2005 Hewlett-Packard

    Development Company, L.P. Page 2

    Contents

    Contents ...................................................................................................................................... 2

    Product introduction ...................................................................................................................... 4

    Purpose of this guide .................................................................................................................. 4

    Product description..................................................................................................................... 5

    Feature Set Overview ............................................................................................................... 13

    W hats new before OVO for W indows 7.5?............................................................................... 40

    Whats new with OVO for Windows 7.5?.................................................................................. 41

    Target market .......................................................................................................................... 50

    Deal assessment......... ....... ........ ........ ........ ........ ........ ........ ........ ........ ....... ........ ........ ........ ........ 51

    Enabling technologies and their buzzwords........ ........ ....... ........ ........ ........ ........ ........ ........ ........ . 52

    Competitive landscape ............................................................................................................. 55

    Versions and supported platforms....... ........ ........ ........ ........ ........ ........ ........ ....... ........ ........ ........ 60

    Requirements........................................................................................................................... 62

    Product integrations....... ........ ........ ........ ........ ........ ........ ........ ........ ....... ........ ........ ........ ........ .... 64

    Additional or related products................................................................................................... 71

    Integration with other non-HP Products and SPIs.... .... .... .... .... .... .... .... ....... .... .... .... .... .... .... .... .... .... 75

    Architecture................................................................................................................................ 76

    Building the quotation ................................................................................................................. 76

    Building Your Own Application Management Solution ........ ........ ........ ........ ........ ........ ........ ........ . 76

    Scalability............................................................................................................................... 79

    Performance considerations fit ........ ........ ........ ........ ........ ........ ........ ........ ....... ........ ........ ........ .... 80

    Order and configuration guide.................................................................................................. 84

    Information required from the customer........ ........ ........ ........ ........ ........ ....... ........ ........ ........ ........ 84

    Ideal technical fit...................................................................................................................... 84

    Demonstration ............................................................................................................................ 84

    Quick Installation guide Reusable Engagement Packages........ ........ ........ ....... ........ ........ ........ .... 85

    Manuals.................................................................................................................................. 85

    Online-help ............................................................................................................................. 85

    Marketing Material .................................................................................................................. 85

  • 8/3/2019 ovowin_75_tsg

    3/91

    Copyright 2005 Hewlett-Packard

    Development Company, L.P. Page 3

    White papers and other technical documentation ....... ........ ........ ........ ........ ........ ........ ........ ........ . 85

    HP-internal............................................................................................................................... 87

    Contacts.................................................................................................................................. 87

    Evaluation.................................................................................................................................. 87

    Where to get the software......................................................................................................... 88

    What to do (and what not to do!) .............................................................................................. 88

    Typical project plan.................................................................................................................. 88

    Who do you need to get it done?.............................................................................................. 88

    Troubleshooting.......................................................................................................................... 88

    Troubleshooting Tools........ ........ ........ ........ ........ ........ ........ ........ ....... ........ ........ ........ ........ ........ 88

    Online troubleshooting tools and features ........ ........ ........ ........ ........ ....... ........ ........ ........ ........ .... 89

    Product configuration tips and techniques........ ........ ........ ........ ........ ....... ........ ........ ........ ........ .... 89

    Dos and Donts....................................................................................................................... 89

    Training..................................................................................................................................... 89

  • 8/3/2019 ovowin_75_tsg

    4/91

    Copyright 2005 Hewlett-Packard

    Development Company, L.P. Page 4

    Product introduction

    Purpose of this guide

    This Technical Sales Guide (TSG) is being created as a way to brief those professionals focused on presales

    technical support about the capabilities, configuration, and performance considerations of HP OpenView

    Operations for W indows and the available SMART-Plug-Ins (SPIs). System management concepts and

    functionality are discussed.

    Throughout the document we use OVO for Windows as abbreviation for the Operations for W indows solution.

    OVO for Unix will be used as abbreviation for Operations for HP-UX/ Solaris.

    This guide assumes that the reader has a basic understanding of network and systems management concepts

    and processes.

    All examples and figures contained in this document are for illustration purposes only. Refer to HP

    OpenView documentation sets on the products listed above for more detailed information.

    Using the Technical Sales Guide, computer professional will be able to:

    Provide accurate technical information regarding solution capabilities.

    Configure basic solution architecture.

    Map customer requirements to technical capabilities.

    Understand the product structure to prepare a quotation for the prospective customer.

    Demonstrate the product solution.

    Set up an evaluation of the product solution.

  • 8/3/2019 ovowin_75_tsg

    5/91

    Copyright 2005 Hewlett-Packard

    Development Company, L.P. Page 5

    Product description

    Managing a multi-vendor environment requires an approach that simplifies the potentially complex problems of

    communication and coordination, and treats the network as a collection of cooperative, communicating

    entities. OVO for W indows is based on the Manager-Agent concept. The agents role is to collect managementdata and send exceptions to the central manager. The manager provides a central repository to centrally store

    these messages for access from operator consoles as well as a central location for the management

    configuration. The link between the manager and the managed nodes is implemented by means of agents,

    which are running on the managed nodes. The manager is split into a management server and a management

    console. The product is implemented using Microsoft technologies such as Microsoft Management Console

    (MMC), Windows Management Instrumentation (WMI), and the Component Object Model (COM).

    OVO for W indows manages the availability and performance of the distributed systems, network elements,

    and applications through policies. By configuring and deploying policies, you define which management

    information should be monitored; for example the Threshold Monitoring policy defines what performance data

    should be collected.

    The performance monitoring is built on the capabilities of the embedded performance component (code name

    CODA). This component is a dynamic data collection and logging component that, based on the instructions

    from a threshold monitoring policy, collects metrics from all measurement sources: program, external, SNMP

    MIB, W MI, and perflib. By default, a set of so-called Golden Metrics will always be collected. Using this

    historical data enables examination of resource utilization and performance trends, the information needed to

    understand how well the IT resources are performing.

  • 8/3/2019 ovowin_75_tsg

    6/91

    Copyright 2005 Hewlett-Packard

    Development Company, L.P. Page 6

    Figure 1: The Big Picture

    The management server provides all possible management functions that can be performed on several hundred

    managed nodes. It centrally deploys policies onto all platforms and agents onto Windows 2000 and NT

    systems. The management server stores messagesand policy configuration and other data to manage services

    throughout the enterprise (for example, services and relationships to other services).

    An intelligent agent on each managed node is responsible for the communication with the management serverand for the execution of corresponding actions on the managed nodes. The agent receives requests from the

    management server to either run or stop a policy, and communicates back on an exception-basis any relevant

    data or messages collected by the policy. The management server and the agents communicate using remote

    procedure calls (RPC), HTTP and DCOM (package deployment only). The agent stores all its performance data

    locally, based on instructions configured via a policy deployed to the managed nodes. This data can be

    viewed graphically for assisting in trend analysis, regular performance reporting, and can be used in

    conjunction with generating events that can be forwarded to the message browser. The agent can also work

    autonomously, which means that whenever the agent cant communicate with the management server (either

    the server or the network link is down) it will locally store the messages and run automatic commands. W hen

    the connection to the Management Server is re-established the locally buffered messages are forwarded to the

    Management Server.

    Management consoles provide a graphical user interface visualizing and editing information stored

    by the Management Server. From the management console you can initiate service and node discovery

    and configuration, policy editing & deployment as well as access runtime data such as messages, service

    maps, and performance graphs. The management console is a Microsoft Management Console (MMC)

    snap-in giving the user a W indows native look and feel. The management console can be installed on

  • 8/3/2019 ovowin_75_tsg

    7/91

    Copyright 2005 Hewlett-Packard

    Development Company, L.P. Page 7

    separate computers and enables multiple operators that can manage the environment from different

    computers. As user roles are applied to the Windows user, each user may have a limited view and

    limited capabili ties thus splitting the operator responsibi lity across the operational staff. Multiple

    Management Consoles can connect concurrently to the Management Server.

    Process Steps

    OVO for W indows provides functionality for performing daily operational tasks and for managing and

    resolving problems to guarantee the availability of the managed environment. Daily operational tasks typically

    include monitoring the managed environment and responding to problems, requiring the following basic steps:

    Identification -observing the environment

    Notification - becoming aware of the problem

    Analysis -understanding the problem and its cause and its impact on the enterprise

    Solution -determining how to solve the problem

    Resolution -starting an automatic or operator-initiated command to resolve the problem

    Tracking - closing and documenting the problem

    For automating these fundamental steps and linking events, recover commands and instructions OVO for

    Windows provides the four process steps: collecting, processing, presenting, and acting.

    Figure 2: Procedural flow on how to integrate management information into OVO for W indows

    Collecting gathers the status information from the managed environment. OVO for W indows Intelligent Agents

    provide real-time monitoring of events that occur within the managed environment and collects performance

    Managed Objects

    1. Collection

    2. Processing

    4. Action3. Presentation

    Automatic

    Actions

  • 8/3/2019 ovowin_75_tsg

    8/91

    Copyright 2005 Hewlett-Packard

    Development Company, L.P. Page 8

    information from services and nodes. Events can originate from various managed objects like networks, web

    stores, systems, databases, monitored applications, or IP devices.

    OVO for W indows supports the following event sources, which are essential in commercial IT environments:

    Perflib

    WMI Store

    Windows Event Logfiles

    System Logfiles

    External Messages (opcmsg command and API)

    SNMP Events

    Custom metrics made available by scripts or external programs

    Another important source of information is the performance data, like Perflib metrics, W MI etc., that OVO for

    Windows can be configured to collect. Collecting performance data continuously on managed nodes can give

    valuable information for use in trend analysis and forecasting, as well as allowing you to analyze performance

    bottlenecks and improve service levels for users.

    Performance data is usually available in abundance; the goal is to reduce this data to important information.

    OVO for W indows assists you in determining what key performance data are relevant, it provides the

    capabilities to monitor several key metrics and thresholds on these key metrics.

    An example of a typical event, in this case a customer variable, is a monitored system parameter, which

    exceeds its defined threshold. N ot all messages sources generate events in the same fashion. OVO for

    Windows provides a standardized message format for all event sources. In general, the event sources

    normalize outside-world data received in their particular format into a common format for further processing

    and presentation to operators.

    On the agent level, OVO for W indows uses filters and thresholds to assure that only relevant information is

    forwarded to the central system, reducing network traffic and providing the scalability required to manage

    large distributed environments. OVO for W indows follows the principle of management by exception, that is, it

    generates messages only when there is trouble. Automatic commands can be configured to start when a

    defined event occurs and bring the status of the managed object back to normal without any manualintervention. The configuration of automatic commands is easily made, from a central point, without any

    programming.

    The OVO for W indows agent is directed by a set of management policies or short policies. The policies hold

    configuration data for the OVO for W indows agent and are stored centrally on the OVO for W indows

    management server. Policies maybe created, modified and deployed to the agent systems. Policies direct the

    agent software what sources are monitored and what events a generated. Each policy in itself may have

  • 8/3/2019 ovowin_75_tsg

    9/91

    Copyright 2005 Hewlett-Packard

    Development Company, L.P. Page 9

    different rules to instruct the agent in detail how to react to a certain state of the managed node. An example

    policy would be a W indows 2000 Application Log Policy monitoring MS Exchange 2000 Server entries in the

    Windows application event log. An example rule within this policy would be to generally forward all critical

    MS Exchange log entries to the management server. If a MS Exchange event meets the condition of a rule, a

    set of customizable actions can be configured. An action is usually a message that is forwarded to the

    management server and optional an automatic or operator-initiated command.

    Processing selects important or critical status information and makes it available on the central system in a

    consolidated fashion.

    All relevant information passing from OVO for W indows intelligent agents to the central management server is

    consolidated and converted into a standard OVO for W indows message format. Although the information is

    extracted from different sources, OVO for Windows assures consistent presentation of information. Regardless

    of the information's source, operators respond using a consistent, intuitive process.

    Unusual conditions (events) will result in a notification of the operator through the Management Console, OVO

    for Windows's MMC graphical user interface.

    Additional message management functionality includes message classification and filtering, and assuring the

    presentation of important information. Positive and negative filtering is provided. Positive filters forward events

    that match a specified pattern for presentation to the operator or start an automatic command. Negative filters

    suppress events matching specified patterns.

    OVO for W indows provides the capability to locally log forwarded or suppressed events that matched a rule

    as well as events that didnt match any rules. OVO for W indows classifies events not matching any filter as

    "unmatched". Unmatched messages often occur with new or undefined sources. Unmatched messages can be

    analyzed to configure appropriate positive or negative filters for smooth event handling if they are likely to

    appear again. Although not required, a good setup of OVO for Windows message sources includes policies

    and rules for all events. This ensures that on any incoming event the appropriate action will be performed,

    regardless of the event source.

    Proactive warning to prevent problems in advance is possible with OVO for W indows by configuring OVO for

    Windows to watch for specific events and collect and analyze performance data, which indicate that a

    problem is very likely to occur within a short time. An example would be to send a notification message if disc

    space has increased with more than 10% in the last hour. Early warning of potential problems enable

    operators to perform pro-active problem management by fixing potential problems before end-user services are

    affected.

    OVO for Windows also allows messages to be combined into logically related groups. A message group

    combines messages from multiple, related sources, and allows sorting them into certain categories.

    Presentation provides different views of the problems that occurred in the managed environment. The

    messages, nodes and services are colored depending on their severity.

  • 8/3/2019 ovowin_75_tsg

    10/91

    Copyright 2005 Hewlett-Packard

    Development Company, L.P. Page 10

    The operators working environment is the Management Console, which gives a view of the service

    environment from a business, organizational, or system perspective, depending on how it is defined by the

    user. The operator manages the service components by using the various views in the OVO for W indows

    console tree and details pane in MMC. The console tree offers a hierarchical, visual representation of the

    nodes, services and tools. The detail pane hosts the list view, message browser view, and map view. The

    message browser is the view of all the active messages received at the console. The map view presents a two-dimensional, graphical view of the entire service or node hierarchy, including any subsystems or sub services.

    The service map allows the operator to drill down to the root cause of a problem and to obtain information on

    how the services within the managed environment are connected and what other services and entities might

    also be affected.

    Figure 3: Example Service Map View

    The map views root cause and impacted services options are excellent tools for a quick identification of the

    problem and its impact, which helps the operator to make the right prioritization decision among the

    problems, as it is possible to easily see which problem impacts the service environment the most.

  • 8/3/2019 ovowin_75_tsg

    11/91

    Copyright 2005 Hewlett-Packard

    Development Company, L.P. Page 11

    Another example shows the OVO for W indows performance grapher:

    Figure 4: Example OVO for W indows Performance Grapher

    The grapher is launched either within the context of a service or node, or underneath new grapher items in the

    MMC console. The grapher connects to the embedded performance component on the agent via a http

    connection. The system has a pre-defined set of graphs included out of the box (e.g. CPU, disk, network

    usage), more graphs are part of the SPI installations. New graphs can be created. The new grapher in OVO

    for W indows 7.0 replaces the PerfView grapher technology used in former versions of the platform. Graphs

    can be used in operator initiated actions.

    OVO for W indows uses colors to indicate the severity of the messages as well as the status of services, node

    groups and nodes. HP OpenView's color coding scheme is based on the OSI telecom standard.

    Figure 5: HP OpenView's event color coding

  • 8/3/2019 ovowin_75_tsg

    12/91

    Copyright 2005 Hewlett-Packard

    Development Company, L.P. Page 12

    The operators use the message browser as a starting point to review and respond to each message. Here they

    find all related message information including the availability and status of all pre-configured automatic

    commands. The operator can also add personal annotations to messages to document what steps were taken

    and to use this information at a later stage to streamline the problem resolving process.

    Furthermore, the OVO for W indows service view provides an overall view of the IT environment and allows

    users to analyze the impact on the overall business services by providing end-to-end service oriented business

    views.

    Acting performs planned activities and automatic and operator-initiated commands, stores information and logs

    the actions (audit).

    When operators use OVO for W indows to manage the nodes in the network, there will be many repetitive

    tasks, which they have to carry out. A set of standard tools comes with OVO for W indows for the most

    common tasks; it is also possible for the administrator to define custom tools.

    OVO for W indows provides the concept of Tools: Operators can perform actions on one or multiple remote

    nodes without the need to physically log on. The OVO Administrator can define the user context and password

    in which the tool is launched, without making this information available to the operator.

    It is possible to create a broad range of tools for the daily operating tasks such as checking the available disk

    space on a node before running or installing a program, or starting management applications, verify if all

    processes (services) are running. The tools may be of five different types; executable commands, Java Script,

    VB Script, W indows Scripting Host Script or W eb URL.

    These tools can also be associated with a service or a node to provide context when the tool is launched and

    to let the operator know what tools are available when operating on a particular service or node.

    The OVO for W indows administrator can associate commands with messages to resolve problems. A

    command can be the execution of a VB script (Windows platform only), a Unix shell script, a program or anapplication There are two types of commands; each message can be associated with one command of each

    type:

    Automatic commands are commands started without operator intervention upon creation of an OVO for

    Windows message. If local automatic commands are configured, these commands are immediately triggered

    by the OVO for W indows intelligent agent, which is hosted on the managed node. This is possible even if the

    central OVO for W indows management server is not running, or if the network is down. The agent will in this

  • 8/3/2019 ovowin_75_tsg

    13/91

    Copyright 2005 Hewlett-Packard

    Development Company, L.P. Page 13

    case log the message and performance data on the managed node, and continue to run automatic commands

    as specified by the policy. W hen the connection to the management server is re-established, the locally

    buffered messages are forwarded to the management server. An example of an automatic command could be

    the deletion of all temporary files on a hard disk if the space exceeds a pre-defined threshold.

    Operator-initiated commands are commands predefined by the administrator, and started by the operator inresponse to a message that indicates a problem. An example of an operator-initiated command might be to

    restart a shutdown application on the operators request, or to start an analysis application, like the

    Performance Grapher for performance analysis, on a system reporting a performance bottleneck.

    Feature Set Overview

    OVO for Window s Management Console

    The OVO for Windows console is composed of a set of views constructing a management environment for the

    operator. A view is a virtual representation of some aspect of your IT environment. Views allow operatorspinpointing important information. Views can be as broad as viewing the entire enterprise IT infrastructure, or

    as specific as seeing only the nodes that are causing a particular service to fail. Here are examples of the types

    of views OVO for Windows provides:

    End-to-end (top down view) of a service or segments of the service chain.

    View of impacted services (if one of the dependent sub services failed).

    Ability to view the root cause of a problem, i.e. the sub services that impact higher-level services.

    View all messages or filtered views of messages for a managed node, a group of nodes, or a service.

    View of tools (the entire tool tree, sub-tree or the tools associated with a service or node)

    Administrative views on managed nodes, e.g. installed agent functionality and policies.

    The following illustrations show the basic elements of the OVO for W indows management console.

    Figure 6: OVO for W indows Management Console

  • 8/3/2019 ovowin_75_tsg

    14/91

    Copyright 2005 Hewlett-Packard

    Development Company, L.P. Page 14

    You can manage a service as a whole, or you can manage the services components by using various views in

    the OVO for W indows console tree and details pane in MMC. The Services folder in the console tree is a

    hierarchical, visual representation of a service, such as printing, e-mail or Internet connection, and itssupporting software components. The managed nodes folder is a visual representation of the managed nodes.

    The details pane on the right side hosts the list view, message browser view, and map view. You can also start

    the Performance Grapher to graph system, network, or application data.

    The OVO for W indows service console allows you to:

    Focus on services that impact your business the most using the map view.

    Display information in your preferred format such as the map view or list view.

    Sort messages in the order you want, according to various criteria, in the message browser.

    Display all managed nodes the management server has "discovered," and all nodes that are being

    managed in OVO for W indows.

    View which policies are installed on a managed node.

    Display the tools associated with a service or node.

  • 8/3/2019 ovowin_75_tsg

    15/91

    Copyright 2005 Hewlett-Packard

    Development Company, L.P. Page 15

    Message Brow ser View

    The message browser is a view of all the active messages received at the console for all nodes and services. It

    is possible to open a message browser at each level of the node and service tree, to see the messages only for

    the selected node or service. Additional the message browser allows the operator to filter on a set of attributes,

    e.g. the message severity or a pattern in the message text etc.

    Figure 7: OVO for W indows Message Browser

    Using the message browser, you can:

    Evaluate all current messages according to color-coded status

    Read message text

    Review instructions and review and add annotations to a message

    Determine impacted services

    Start operator-initiated commands

    Review status of operator-initiated commands

    You can simplify the message display by showing only messages from systems or services of special

    interest to you. The browser headline displays the message attributes in a condensed format you can

    interpret quickly.

    Service M ap

    The service map is an integral part of OVO for W indows and helps IT operators to easily focus on the business

    services. The overall view of the IT environment that it provides allows operators to analyze an event in terms ofimpact on the overall business services. Operators can see at glance the operational status of the service

    environment. The service view also provides a way to describe the dependencies between service, hosts (incl.

    the associated hardware) and processes running.

    Below is an example of a service map. The Map View shows the entire service hierarchy, the type of

    relationships the service has with other components/ services, and nodes supporting that service. You can also

  • 8/3/2019 ovowin_75_tsg

    16/91

    Copyright 2005 Hewlett-Packard

    Development Company, L.P. Page 16

    view just a specific sub service or node. The services are colored to reflect their status, just as with messages.

    For each service you can open a message browser containing the messages related to the service.

    The service map view features a Hyperbolic Tree that enables navigating in larger service maps by

    automatically centering and enlarging a highlighted service icon. While navigating through the service tree, the

    Hyperbolic Tree unfolds depended or related sub services.

    Figure 8: OVO for W indows Management Console: Service Map View

    To manage the services more efficiently, the Service Console comes with the following features:

    A root cause view, which displays all the services that have led to the current status of the service. This

    helps the operator to quickly identify the core problem within the service hierarchy and the related

    messages in the message browser.

    An impacted services view, which identifies at a glance the higher-level services that are affected by a

    problem in a certain sub service.

    Service specific tools can be defined for each service in the hierarchy by the OVO for W indows

    administrator. The tools can be run on a service from the launch tool menu.

  • 8/3/2019 ovowin_75_tsg

    17/91

    Copyright 2005 Hewlett-Packard

    Development Company, L.P. Page 17

    Service configuration, which is made easy through the service configuration editor.

    Within the service hierarchy each service has a Service ID. OVO for W indows messages have a reference to

    this Service ID to feed real-time data into the service map. The state of a parent service is defined through status

    calculation and status propagation from its child services and/ or through messages for the service itself.

    Service hierarchies can be based on any kind of relationship such as on applications, geography,

    organizations and infrastructure. Operators can have different entry points in this service hierarchy so that each

    operator can focus on the services in his particular area of expertise and responsibility.

    When a message with a specific service ID is received, the status of the entire service will be evaluated

    according to the status calculation rules that have been established for the particular service, which is affected.

    Service Conf igur ation

    The service editor allows you defining how services in the Service Hierarchy are dependent on each other, and

    allows you defining rules that evaluate the severity based on the state of the contributing services. Although youcan use the service editor to create an entire service hierarchy, SMART Plug-Ins -or short SPIs - (from HP or

    other vendors) provide the capability of inserting a basic hierarchy, which you can then fine-tune and connect

    to higher levels within your service hierarchy. SPIs such as the SMART Plug-In for W indows, contain policies

    and an auto-discovery feature that automatically discovers managed nodes and processes running on the

    node, assign and deploy automatically matching policies, and create a basic service hierarchy. The discovered

    services hierarchy then can be modified by using the service editor to match the requirements for your

    environment. You can also build on the SPI service hierarchy, using it as a building block in a larger hierarchy

    that you create.

    NOTE: The auto-discovery, auto-assignment of policies and the automatic creation of the service hierarchy are

    features provided by SMART Plug-Ins.

    If you plan to create your own service hierarchy from scratch, or do a major extension of the existing map, it is

    helpful to draft your service hierarchy before you start using the service editor. When planning your service

    hierarchy, keep the following questions in mind:

    Which IT services do you provide? Which ones do you want to monitor?

    Who are the customers of your services? Which organizations, departments, or lines of

    business are important to you?

    How can you logically group the services you provide? Which services are used by other

    services?

    How do problems in one service affect another? Which status propagation rule should you

    apply?

    How do you evaluate the severity of a problem? W hich status calculation rule should you

    apply?

    Which service actions are appropriate?

  • 8/3/2019 ovowin_75_tsg

    18/91

    Copyright 2005 Hewlett-Packard

    Development Company, L.P. Page 18

    Determination of status severity depends on the rules set up in the service configuration editor. For example, if a

    service is comprised of five other services, each of which receives a "warning" message, what should the final

    status be? Warning? Critical?

    There are different approaches you can use to determine the severity of a service:

    Status Calcula tion defines how to compute the severity status of a service from the status

    severity levels its sub services propagate. Basically, the services own status is calculated from

    the status of all the contributing sub services, and messages affecting the service itself.

    Status Prop ag ation defines whether a sub service is more or less important compared to

    other sub services. It describes how the status of a child service is interpreted by the parent

    service.

    Severity calculation must often account for many variables, it is recommended to look at existing rules provided

    by SPIs for examples of status calculation and propagation.

    How events and messages map to serv ices

    In previous chapters, we explained that events trigger messages, and that messages provide status information

    on a managed node in the OVO for W indows console. Since services are high-level abstractions that must be

    defined within the IT environment, it is necessary to establish a link between individual services, service

    hierarchies, and incoming messages. The management server maps a message to a service by its service id. To

    ensure correct mapping of messages to service(s), you must use a unique service id for each service and use it

    consistently in the service editor and in the configuration of the policies outgoing messages.

    Ex am ple of M anag ing a Business Service

    THE SCENARIO: You want one operator to monitor your accounting service, which includes general ledger,

    accounts payable, and accounts receivable software programs as well as databases across several managed

    nodes, as shown below: Here, we assume that you will use default values (for example, propagation rules,

    messages) for the services and policies you establish.

    1.Draw a simple diagram of the different elements you want to monitor for the accounting service, and

    determine the dependencies. An example diagram is shown. The dotted lines indicate dependencies, which

    you establish later.

    Figure 9: Service View dependencies examples

  • 8/3/2019 ovowin_75_tsg

    19/91

    Copyright 2005 Hewlett-Packard

    Development Company, L.P. Page 19

    2.Make the nodes where the accounting applications elements (databases, programs) are located managed

    nodes, launch the NT Service Discovery if the node is a Windows node to build the System Infrastructure

    services, and the Windows OS policies will be deployed automatically.

    3.Add the component services, which are the building blocks of your accounting service model. For example,

    you would add the General Ledger component, and indicate that it is hosted on managed node A.

    Create dependencies between services. For example the databases are more than likely a component of

    another service (e.g. Oracle or SQL Server), and thus the accounting services would be dependent upon thesedatabase services. You may also want to build dependencies upon different System Infrastructure services to

    identify different types of problems.

    Create additional policies to monitor and associate messages with the services. (See online help for

    information on message attributes).

    The Operator tasks:

    1. If one of the monitored services changes state to something other than normal, perform a root

    cause analysis to determine which service caused the abnormal state2. If the service is in the operators control, retrieve the active messages and select the message that

    caused the problem

    3. Own the message

    4. You may want to do an impact analysis to determine the effect this component has on other

    services in the enterprise (this is helpful in prioritizing multiple problems)

  • 8/3/2019 ovowin_75_tsg

    20/91

    Copyright 2005 Hewlett-Packard

    Development Company, L.P. Page 20

    5. Review the message and instruction texts to get additional information about the problem

    6. Perform an operator-initiated action if one has been established by the administrator. This may

    resolve the problem or give additional information to help resolve the problem

    7. Annotate the message to indicate how the problem was solved

    8. Acknowledge the message so it moves to the Acknowledged Messages browser

    3.5 Collecting Mana gement Inform ation

    On each managed node, OVO for Windows establishes various message collection services (OVO for

    Windows intelligent agent) to gather exactly the management information the operators need to get their job

    done. These services extract and initially classify the information by assigning values to the message attributes

    severity, service name, application name, message group, object name and managed node name, and others.

    Messages can be captured from various sources described below, regardless of whether the event is related to

    network, system or applications specific incidents. If the original information text is cryptic (that is in binary

    format) or a number (for instance an SNMP object identifier), it is possible to map the original information to a

    readable message. Parts of the original text (for instance the error number) can be used in the message, which

    is forwarded to the operator.

    Windows Security, Application and System events are written to the Windows Event Log and can be captured

    by the OVO for W indows Windows Event Log policy. Some Windows Event Log messages have pre-defined

    attributes provided by the application sending the message. These attributes are automatically mapped to the

    OVO for W indows message.

    System/ Application Logfiles contain events written by applications to communicate their status and error

    information. The Logfile Entry policy analyzes logfiles, extracting the information and assigning the configuredattribute values to all messages extracted from this source. Original logfile entries can be modified with a

    powerful pattern matching language (similar to regular expressions) to build new more meaningful message

    text. Preprocessing tasks performed by the OVO for W indows intelligent agent locally on the managed node

    can be defined for the logfile, for example, to convert binary logfiles to an ASCII representation.

    The OVO for Windows Message Application Programming Interface (API) and Command Line Interface (CLI)

    allow applications to directly feed status and/ or event information into OVO for Windows on the managed

    node via the OVO for W indows Message Interface. Typically, the API or CLI will be used within new

    applications or user defined scripts to deliver the information most efficiently to the management server.

    Monitoring Services

    Sophisticated monitoring is key to providing exception-based and pro-active problem resolution. The OVO for

    Windows agent provides distributed monitoring capabilities. With the Measurement Threshold policy

    functionality it is possible to define threshold conditions, which the monitored values periodically are checked

    against. If the thresholding condition is met, a message is generated to inform the responsible OVO for

    Windows operator. It is possible to define several threshold values for the same source, for example to

  • 8/3/2019 ovowin_75_tsg

    21/91

    Copyright 2005 Hewlett-Packard

    Development Company, L.P. Page 21

    configure the policy to send a warning message when disk space of 90% and a critical message when it is

    95% full.

    Monitoring values can be gathered from any program or user defined script, Perflib, the embedded

    performance agent, as well as retrieved from SNMP Management Information Bases.

    Custom monitoring applications can make use of the monitoring API and CLI to feed their values directly into

    OVO for W indows. They can either be triggered by OVO for W indows to deliver their values or the

    information is asynchronously sent from the monitor.

    Minimum or maximum threshold values as well as Visual Basic Scripting expressions can be used on the NT

    managed node to combine multiple metrics in a condition or to calculate values from multiple metrics and

    threshold the results.

    The use of VBScripting allows the user to do advanced calculations of a threshold or multiple thresholds, for

    example calculating a threshold value for multiple monitor sources, such as a threshold for the total free disc

    space for all discs (C, D, and E) on a computer. Another example is to create a critical alarm if the usedprocessor time is greater than 90% and the processor queue length is greater than 3.

    Messages generated by OVO for W indowss monitoring service consist of the message attributes and the

    definable message text, which can include actual monitoring values. The messages are treated identically to

    those of other sources, so that logging options and message bound actions can be applied.

    Figure 10: Example for objects monitored by default

    M onitored Object Threshold Reset Va lue Monitor ing

    Value

    CPU utilization 90% 85% 5 min.

    ProcessorQueueLength 3 2 5 min.

    3.6 Using Policies to M anag e Systems

    The concept of OVO for W indows management policies allows you to centrally define the rules on how to

    monitor objects throughout your IT environment. For example, the W indows Event Log policies allow you to

    intercept Windows Event Log messages matching defined conditions based on the rules created or modified by

    the OVO for W indows administrator. Policies then can be distributed to one or multiple managed nodes. OVO

    for W indows provides versioning on policies and inventory reports for managed environments. A sample

  • 8/3/2019 ovowin_75_tsg

    22/91

    Copyright 2005 Hewlett-Packard

    Development Company, L.P. Page 22

    inventory report would include the managed nodes and currently active policies along with their version

    number. If a specific condition is met on a managed node, a message is generated and sent to the OVO for

    Windows management server and displayed on one or more OVO for W indows console systems. This

    concept allows you to:

    Consolidate event messages from all computers to a central location Handle events on a priority basis (e.g. by severity)

    Add annotations to a message to document steps taken

    Execute operator-initiated commands

    A policy is a set of rules and actions that help automate network and system administration. Using OVO for

    Windows, the administrator can deploy policies to various destinations to provide consistent, automated

    administration across the network.

    Every policy belongs to one policy type. Each policy type has its own editor on the OVO for W indows

    administrators console.

    OVO for W indows is an open platform: The design of OVO for W indows allows to introduce new policy

    types along with the associated policy editor and new agent or subagent functionality.

    Overv iew on Policy Typ es

    Logfile Entry: Evaluates (binary or text) logfi les on the node (UX and - less common on - NT), andgenerates messages that match a rule in this policy.

    Open Message Interface: Intercepts messages that are externally generated by the opcmsgcommand or API call (UX and NT). It compares a message against rules defined in an Open Message

    Interface Policy. Rules process the information provided in the command line or API call further, e.g.

    automatic or operator-initiated commands are associated and instructions attached.

    Windows Event Log: Monitors the NT Security, Application, System and other Event Log messagesand generates messages for matching entries.

    SNMP Interceptor: Intercepts SNMP traps on managed node (for UX and NT) and generatesmessages for matching the rules.

    WMI Source: Provides monitoring of W MI instances (e.g. application data) with a sophisticated setof rules and features (Windows only).

    Scheduled Command: Runs a specified command (UX and NT) according to a specific schedule,intercepts the results and sends a message if the command was successful or not. Recently, scheduled

    command policies also support execution of VB Scripts (Windows only) and Perl scripts. A number of

  • 8/3/2019 ovowin_75_tsg

    23/91

    Copyright 2005 Hewlett-Packard

    Development Company, L.P. Page 23

    COM objects are available to trigger actions such as sending a message to the console or collecting a

    metric in the embedded performance agent.

    Serv ice Process Monitor : Added for ease-of-use, user friendly selection of services and processes to

    be monitored. Implemented in the threshold monitoring policy, it uses VBScript for service monitoring

    and Perl for process monitoring. New with OVO for W indows 7.5

    Threshold Monitoring Policies:

    Measurement Threshold: Polls different measurement sources (for example real time performance

    monitoring = perflib), determines whether the values exceed configurable parameters, and sends

    notification messages to the message browser. The Measurement Threshold policy has access to

    various types of sources, e.g. Perflib (Windows only) or MIB variables (local or remote node). Other

    sources are program/ scripts submitting opcmon values (commandline or API) and the type external that

    - similar to the Open Message Interface -evaluates values submitted by an external application.

    On the Windows platform the Measurement Threshold Policy can utilize scripting (via Visual Basic) for

    sophisticated multi-source thresholding.

    Performance Collection Policies:

    Performance Collection Policies: These are implemented via Measurement Threshold policies and are

    combined with thresholding. You can collect metrics based on all sources for measurement threshold

    monitors. If you simply want to collect metrics, just leave the threshold tab empty.

    Flex ible Ma nagem ent Policies: These are policies supporting Manager of Manager (MoM)

    configurations, also named flexible management configurations. Note, only one policy of this

    type is active, so the last policy deployed wins.

    NodeInfo Policies: These allow central configuration and maintenance of the nodeinfo file. You canspecify attributes such as port ranges, health checking and tracing with this configuration file.

    Discovery Policies: These are policies used to trigger automatic discovery of services at a given

    polling interval. After the initial discovery is completed, only the deltas will be forwarded to the

    management server and made available to the console users. With former versions, discovery

    and synchronization was offered via the means of tools.

  • 8/3/2019 ovowin_75_tsg

    24/91

    Copyright 2005 Hewlett-Packard

    Development Company, L.P. Page 24

    Figure 11: Deployment Packages

    Policy Deployment

    Deployment Packages contain software and intelligent agent components to be deployed on the managed

    nodes. OVO for Windows has a single main agent components on the Windows platform:

    OpenView Operations Agent: This includes the formerly known Enterprise Message/ Action Agent and the

    embedded performance component

    OpenView Operations Agent is the core component on the managed node. The three key parts are the

    Message Agent, which is handling the message forwarding to the management server, the Action Agent,

    which is launching automatic and operator-initiated commands, as well as tools (initiated from consoles), and

    the Embedded Performance Component. The key function of the performance component is polling of metrics

    in a defined time interval and storing the collected data locally on the managed node. The collected data can

    then be accessed via the performance grapher or collected centrally with Reporting tools like the reporter.

    Packages containManageme

  • 8/3/2019 ovowin_75_tsg

    25/91

    Copyright 2005 Hewlett-Packard

    Development Company, L.P. Page 25

    In addition it consists of several subagent process managing various message sources, e.g. for Threshold

    monitoring, Logfile, SNMP and W MI.

    SPI packages are usually optional packages which hold additional functionality that has to be deployed to a

    managed node where the SPI is installed.

    NOTE: A SPI extends the base functionality of OVO for W indows. The term SPI (SMART Plug-In) is typically

    used if the component is an optional add-on to the core OVO for W indows product and has to be purchased

    separately (e.g. SMART Plug-In for Microsoft Exchange Server). Note, there are 3 exceptions of SPIs that

    provide basic monitoring and are free of charge and part of the foundation:

    SPI for web server (formerly known as passive monitoring component (PMC) for internet services

    SPI for Microsoft Windows

    SPI for HP-UX, Solaris, Linux, AIX and Tru64

    Instrumentation packages are additional software components adding a set of executables, libraries and

    configuration files to extend the capabilities of the agent. The tools located in a previously named Windows

    Module Tools package are now modelled as an instrumentation package. Deployment of instrumentation

    gives you the choice to pick individual components. In case you assign policies to policy categories, and you

    deploy one of those policies. the corresponding instrumentation package will implicitly be deployed.

    To deploy policies the administrator has two options:

    Automatic deployment: Functionality provided by SPIs/ Modules that support auto-discovery and auto-

    deployment (Windows only). Auto-deployment might not be desired in large IT environments

    (scalability/ bandwidth considerations)

    Manual deployment: The administrator manually selects and deploys the policies to managed nodes.

    On deploying policies to a managed node, the appropriate agent package - if not already present on the

    managed node - will automatically be deployed along with the policy. Each policy has a dependency to one

    or more agent packages.

    A new feature is synchronizing the nodes policy inventory with the policy inventory that exists on the

    management server. This is useful, e.g. if the node has a pre-installed agent. Then, the synchronization feature

    will detect the existence of this agent, will update the servers policy inventory. Synchronization is implicitly

    called if you want to deploy a policy, or if you explicitly want to synchronize.

    Policy Versioning

    OVO for Windows ships a set of pre-configured (out-of-the-box) policies which cover the most common

    management tasks. These policies can be modified or used as templates for custom policies to meet your

    specific needs. You can also create a new, blank policy from a chosen policy type. The policy editors provide

  • 8/3/2019 ovowin_75_tsg

    26/91

    Copyright 2005 Hewlett-Packard

    Development Company, L.P. Page 26

    support and online help to support you in creating your own custom policy. Every time a policy is modified, the

    minor version number is automatically increased.

    OVO for W indows has two policy management areas:

    The first policy management area is similar to a file/ directory structure: Policies and policy folders can becreated, policies can be grouped in any form. Examples are by node or group of nodes, by platform type, by

    logical function in the IT environment, or by application hosted on the managed node. This area is the typical

    workplace for the administrator.

    In the second area the policies are grouped by type, i.e. each folder holds all available policies of a given

    type. These folders do not allow substructures. The Policies Grouped by Type area is usually used for a

    reference list of all available policies, administrators may copy a policy from this area to the above mentioned

    work area to perform customizations. This area also provides access to previous version of a given policy.

    To obtain an overview over managed nodes and the installed and active policies, as well as an overview over

    the configured services and the used propagation rules, OVO for W indows offers a reporting in HTML format.The reporting tools are available via: Tools->Reporting-> Inventory & Service Reporting.

    Configuration store and transfer: OVO for W indows uses a RDBMS to store and retrieve policies, including all

    versions of policies. To move or transfer configuration data between OVO for W indows management server a

    download and upload tool is available.

    Policy Groups

    With several hundred policies deployed to a large number of managed nodes, keeping track of the installed

    policy is a challenging task.

    Use OVO for W indows features for keeping track of your developed and deployed policies. The versioning

    feature and the ability to create your own hierarchy of policy groups assist you in keeping a good overview.

    A policy group is a logical group, similar to a file system directory. A policy copied into a folder will remain

    the same until it is deleted or replaced by a policy of a higher version.

    The following shows some ideas of how to group policies:

    By platform type (Windows 2000, N T, HP-UX, Solaris)

    Groups for policy development and production

    By location, site

    By application, e.g. MS Exchange server, MS SQL server

    By nodes or node groups

    Policy groups also offer the following advantages:

    Only policies that are members of policy groups can be exported / downloaded

  • 8/3/2019 ovowin_75_tsg

    27/91

    Copyright 2005 Hewlett-Packard

    Development Company, L.P. Page 27

    Policy groups can be assigned to node groups. As new nodes are added to node groups, all

    policies of the assigned policy groups will be deployed. The implicit deployment can be

    disabled.

    Policy Catego ries

    Policy categories is a means of grouping policies and adding a some semantics to them:

    If policies of a certain policy category will be deployed, and an instrumentation directory

    with the same name exists, policy management will implicitly deploy corresponding contents

    of the instrumentation directory

    Policy categories can be used with user roles. A role can be configured to allow permissions on policies only

    for a certain set of policy categories

    3.7 Message Processing

    Agents monitor managed nodes for specific events depending on the policies installed on the managed node.The Event/ Message policy types can generate a message or an action in response to an event. For example,

    monitoring and responding to SNMP events requires an SNMP Interceptor policy to be deployed to that

    managed node. For each event source the OVO for W indows administrator sets up one or more policies with

    positive and negative filters to accept important or suppress irrelevant events.

    When an event occurs, the agent checks if the event matches any of the rules defined for the policy. The rule

    configuration defines specific event attributes including the text pattern the event should match against, and can

    be fine-tuned for each managed node and information source. When an event matches all defined attributes, it

    is qualified for further processing for example forward a specific message to the management server, or start

    local automatic commands immediately.

    The processing of rules in a policy is top down; the event is sequentially processed against the rules. The first

    rule matching generates the message, and the processing is stopped after the first match.

    The order of rules is important: The top down processing order requires the more special rules to be processed

    before more generic rule.

    Example: A disk space policies contains 3 rules, the rules are as follows:

    Critical if disk is used over 95%

    Major if disk is used over 90%

    Warning if disk is used over 70%

    To implement this correctly, the order has to be 1st rule critical, 2nd rule major, 3rd rule Warning. Note that if

    the Warning rule would be the first rule, the result would be a Warning condition on any disk space usage

    over 70% even if it were 99%. In this case the more critical rules would never be reached since the Warning

    rule will already match.

  • 8/3/2019 ovowin_75_tsg

    28/91

    Copyright 2005 Hewlett-Packard

    Development Company, L.P. Page 28

    Suppress Rules: OVO for Windows also offers the concept of suppressing rules. At the first glance a

    suppressing rules seem equal a rule that just doesn't match, i.e. just leave a matching rule out. Why do we

    need suppressing rules?

    Following example of an advanced disk space policy will provide an example for a suppress rule:

    Disk "C:" critical if over 80% used, no events if below 80%

    All disks critical if over 95% used

    All disks major if over 90% used

    All disks Warning if over 70% used

    Assuming now that "C:" disk is used 75%, this would not match the first rule -since it is below 80%, but would

    be matched by the generic "all disks warning over 70%". However this is not desired, since the all events for

    "C:" below 80% should be suppressed.

    To accomplish this insert a suppressing rule

    Disk "C:" critical if over 80% used, no events if below 80%

    Suppress all other Disk "C:" conditions

    All disks critical if over 95% used

    All disks major if over 90% used

    All disks Warning if over 70% used

    Another usage for the suppress rules is within policies that have a large number of rules, e.g. the SNMP policy.

    The suppress rule here can be used to improve the performance: If the SNMP policy is processing only a

    specific pattern in the OID, all events that don't have this pattern can be suppressed before all the otherconditions are being processed and thus saving processing time.

    Unmatched events: Several policy types offer the option to create an event if it does not meet any of the defined

    matches or suppress conditions. There are 3 options what to do with an unmatched event: Forward it as a

    regular event to the active message browser, forward it directly to the acknowledged message browser, or

    ignore the message. The option to forward is usually use in the development stage of a policy. The

    administrator here can see what events are generated by a message source and to create a matching rule for

    events of interest. It is usually best practice to provide a rule for all-important events and turn on the "ignore

    unmatched" switch. For some policy types that receive the events from an external source, e.g. the Open

    Message Interface Policy, is might still be useful to have unmatched events forwarded to the active message

    browser.

    Local logging of events is provided. You can define which events, if any, should be logged on the managed

    node from which they originated, for example by choosing to log events that match a rule and trigger a

    message or are ignored or to log events that do not match any rules in the policy.

  • 8/3/2019 ovowin_75_tsg

    29/91

    Copyright 2005 Hewlett-Packard

    Development Company, L.P. Page 29

    Message buffering on the managed nodes will be activated by OVO for W indows if the management server is

    not reachable. Buffering continues until the connection can be re-established. Local automatic commands are

    still run even if the connection to the management server is down.

    Message specific instruction texts can be linked in OVO for W indows to each message. Specific Instruction text

    may be attached to each rule within a policy, or generally to the policy. They may contain addition instructionsor help for the operator of what has happened and how to best resolve the problem.

    Message specific annotations can be added to each message. An operator usually adds an annotation to a

    message to inform other operators or the administrator about which actions he has taken on the message. All

    annotations are stored in the central repository. Automatic and operator-initiated commands can also be

    configured to add an annotation to the message if completion was successful and the results. In addition the

    annotation shows which operator executed the command, along with start/ stop time. In case of a command

    failure an annotation will always be written.

    The Measurement Threshold Policy features 3 levels of messages, each with individual text and associated

    commands. The Start message is generated if the threshold is violated for the first time, the Continue message isgenerated if the threshold is still violated in the 2nd and all subsequent polling intervals. An End message can

    be configured if the threshold was at least once violated and is now back to normal. To each of these

    messages a Message Key can be associated and can be referenced by other message to automatically

    acknowledge these messages. E.g. an End message can automatically acknowledge all Start and Continue

    message from the same message source, the messages can be correlated.

    Some message sources might generate a lot of duplicated messages (e.g. SNMP traps). These messages may

    fill the Message Browser and don't contain additional of information, and may therefore not desired. Several

    policy types provide a way to suppress duplicate messages. The suppression of duplicates of duplicate

    messages uses either a time interval, for instance suppressing all events occurring more frequently than each 30

    seconds for no longer than 5 minutes, or a counter, for instance accept each 5 th event. It is also possible to

    combine the two methods.

    Forwarding messages to external services such as trouble ticket systems is possible by configuring a WMI

    policy and setting up an automatic command using VB scripting to feed the message information into the

    trouble ticket interface. For further details, see a corresponding white paper.

    3.8 Performance Monitoring on W indow s

    The performance and availability management is based on the same concept of using policies to define which

    data to monitor and collect. The performance data is stored on the managed node and provides a near real-time status as well as an historical view on the service infrastructure.

    The performance monitoring is built on the capabilities of the embedded performance component. This is a

    data collection and logging component. Deploying Threshold Monitoring policies to Windows managed nodes

    start the collection of comprehensive resource and performance data (Perflib, WMI, etc.) from applications,

    networks, and operating systems.

  • 8/3/2019 ovowin_75_tsg

    30/91

    Copyright 2005 Hewlett-Packard

    Development Company, L.P. Page 30

    You can manage the performance data by enabling further metrics to be collected. You do this with the

    help of threshold monitoring policies. If you dont want to collect any metrics you may stop the

    embedded performance agent via the opcagt command, that allows to stop individual subagents:

    opcagt stop id .

    The collection metrics can be accessed via the advanced thresholding monitor or graphed with thePerformance Grapher that can be started from the OVO for W indows console

    Using the historical performance data, the Performance Grapher enables examination and diagnosis of

    resource utilization and performance trends. With this understanding bottlenecks in the IT infrastructure can be

    discovered and eliminated.

    OVO for W indows delivers a standard measurement collection, collecting the golden metrics. This collection is

    available cross platform and will offer a good out-of-the-box solution for most users. For advanced users

    threshold monitoring policies can be used to enable further collections.

    Alarming is configured with the Measurement Thresholding policy:

    The basic Measurement Threshold is a set of rules for typically one source (e.g. a single metric). This single

    source may have additional attributes attached that can be used with the filter, however the source submits only

    a single value.

    An example for a single source would be mentioned "Free Disk Space" policy, it submits a single value (free

    space in bytes) and may attach attributes such as the disk identifier (e.g. "C" or "D") if multiple instances are

    available on the managed node.

    The advanced features of the Measurement Thresholding policy make it possible to create advanced threshold

    conditions for multiple sources, i.e. calculate/ correlate a threshold condition from multiple sources within thesame policy. The single fixed threshold condition from the basic threshold can be extended to multiple sources

    within the same context (policy) and to define the threshold conditions programmatically including multiple

    sources (VBScript or Perl script based).

    These advanced thresholding capabilities provide a very flexible and powerful set of tools to address even

    complex monitoring tasks.

    How to use ARM w ith OVO for W indow s?

    Management of ARM instrumented applications is only possible with the HP OpenView Performance Agent

    (formerly: Measureware agent). Note, you also need the HP OpenView Performance Manager in order to view

    the graphs based on data collected with the HP OV Performance Agent.

    Figure 12: OVO Performance Collection overview

  • 8/3/2019 ovowin_75_tsg

    31/91

    Copyright 2005 Hewlett-Packard

    Development Company, L.P. Page 31

    Description of how OVO for W indows collects and processes performance data

    1 The OVO for Windows central management server (accessed via a local or

    remote console) is the starting point, all policies are centrally stored and

    administered here.

    2 OVO for Windows monitoring policies are created and modified using the

    threshold monitoring policy editor.

    3 The OVO for W indows administrator assigns policies to managed node and

    deploys OVO for W indows components, both policies and packages.

    4 As soon as the common agent, the OpenView Operations Agent is deployed,

    the embedded performance component will start collecting the golden metrics.

    As new threshold monitoring policies are deployed that enable collections, the

    performance component will increase the number metrics collected.

    5 The metrics are sampled and stored at regular intervals to create a historical

    view over time. The historical data can be used to examine resource

  • 8/3/2019 ovowin_75_tsg

    32/91

    Copyright 2005 Hewlett-Packard

    Development Company, L.P. Page 32

    utilization, performance trends, and capacity planning. (See step 10.)

    6 When specified thresholds are exceeded, the Measurement Threshold policy

    instructs the agent to create a message and send it to the management

    server.

    7 The management server forwards the message to all management consoles.8 The message is received, and operators view it in the active message

    browser on the management consoles.

    9 The message contains information and optional operator-initiated commands

    to evaluate performance issues. These operator-initiated commands include

    launching of the performance grapher.

    10 Operators can use the performance grapher to graph and analyze the

    performance data, (the historical data that was collected during the time the

    threshold violation occurred.)

    Perform ance Monitor ing on UNIX Systems

    Performance management of HP-UX and other UNIX systems is possible from OVO for W indows if the

    performance sub agent (=Coda) is explicitly deployed. This performance data from UNIX systems can also be

    accessed using the performance grapher. Alarming is based on the definitions in corresponding threshold

    monitoring policies.

    In addition the HP OpenView Performance Agent (formerly: Measureware) is offered. This gathers performance

    statistics directly from the UNIX kernel into circular, configurable log files. The correlation of data from the

    different logfiles is done automatically by the HP OpenView performance agent processes running on the Unixsystem. This performance data can not be accessed using the built-in performance grapher you need to use

    the HP OpenView Performance Manager to access data from this agent. The OVO for W indows OpenView

    Performance Manager is available shortly after release of OVO for W indows 7.0. Alarming on UX systems

    based on the HP OpenView Performance Agent can be done through the alarmdef file.

    3.9 The Performance Grapher

    OVO for W indows provides a graphical tool to analyze resource utilization information collected by OVO for

    Windows managed nodes. The Performance Grapher can, for example, be used to analyze and document

    long-term IT resource utilization data from one central location.

    Figure 13: OVO Performance Grapher

  • 8/3/2019 ovowin_75_tsg

    33/91

    Copyright 2005 Hewlett-Packard

    Development Company, L.P. Page 33

    The Performance Grapher provides real time and history data, for example, on:

    Health on transactions

    CPU performance

    Memory usage

    Network performance

    3.10 User Roles

    Let me start by having a look at tasks of typical administrators:

    Plan the product implementation

    Install and configure the product

    Maintain the product configuration

    Develop and deploy policies and agents as necessary

    Create tools

    Create and modify service hierarchy

    Tasks of typical operators include the following:

    Monitor the status of a service or system using default or custom views

  • 8/3/2019 ovowin_75_tsg

    34/91

    Copyright 2005 Hewlett-Packard

    Development Company, L.P. Page 34

    Troubleshoot the problem or isolate it using the appropriate combination of filters to research

    the problem or the appropriate view to drill down to it

    Resolve the problem by using tools (an application or program that is not associated with the

    event) or a pre-configured command (script, application)

    Draw performance graphs regularly to verify system performance trends, etc.

    Detect services that have problems or are unavailable, determine services that cause the

    problem and to find the related messages

    Review messages to monitor the health of the managed systems

    Select from a variety of tools, information sources, and problem-solving techniques to resolve

    problems

    Maintain history for all problems you resolve

    OVO for Windows fully supports user roles that can be defined very granular. A given Windows user is

    assigned multiple roles. You can set-up roles for operators, administrators, and a mix of both. Multiple

    concurrent administrator users are supported.

    You can use different users and roles in order to split responsibility in your environment. The responsibility split

    may be overlapping or fully disjunct, depending on the needs of your staffing.

    User roles do not necessarily correspond to separate people in all circumstances, but instead refer to the kind of

    tasks to be performed. In some environments, one person may perform all functions, while in others, the

    operator as well as the administrator role may be divided among several people.

    First, you define the role. A role encompasses viewing rights to services, nodes and tools and capabilities. The

    following are the properties you can use when setting up a role:

    Limited view to services

    Limited view to node groups and nodes

    Limited view to tools

    For messages, you define per message group, whether or not the messages can be viewed, acknowledged,owned, severity changed, and performed other actions on them

    For policy management, you define per policy group, whether the policies contained therein may be viewed,

    edited, deleted, and deployed

    Finally, you assign the role to one or more Windows users. The same is valid the other way round: A Windows

    user may have one or more roles assigned to him.

  • 8/3/2019 ovowin_75_tsg

    35/91

    Copyright 2005 Hewlett-Packard

    Development Company, L.P. Page 35

    However, you need to make sure the Windows users are members of the pre-installed group HP-OVE-

    Operators.

  • 8/3/2019 ovowin_75_tsg

    36/91

    Copyright 2005 Hewlett-Packard

    Development Company, L.P. Page 36

    3.12 Self-Management

    The Self Manager module

    OVO for W indows is an application as any other application on your system and can by monitored by the

    agent installed on the management server; OVO has some self-management capabilities built-in. A service

    model will be part of the Self Manager. The model instantiation for every Windows managed node is done via

    a tool (Synchronize Agent Services Tool) that checks that services from the OVO for Windows agent software

    are installed on the managed nodes and then updates the service tree accordingly. The tool is automatically

    run upon installation and after that scheduled once a day with the help of a scheduled command policy. The

    tool can also be started manually. The set of policies configured to monitor the OVO for W indows related

    events, processes and services are automatically deployed.

    Self-management Policies:

    Windows Eventlog policies configured to monitor OVO for W indows related NT EventLogentries on the management server or managed node. The policies generate messages and

    update the status of related OVO for W indows services in the service map.

    Advanced Thresholding policies monitor the size of the OVO for Windows data repositories on

    the management server and agent and create an alarm if the defined size is over the specified

    threshold.

    Process Monitors monitor all processes/ services that always have to be running on the OVO

    for W indows management server and managed node. This is implemented using the VB Script

    capabilities of the Advanced Thresholding policy.

    Scheduled Command Policy is used for the resynchronization of the services.

    Database Management is provided on the management server and managed nodes to make sure that the

    databases do not overflow. History messages older than (by default) 30 days are removed from the database

    on the management server (this setting is customizable). On the managed nodes, setting collections at specific

    sizes helps to maintain the size of the database.

    Heartbeat polling is provided to detect any problems on the managed nodes. It is checked that the intelligent

    agent on the managed node is correctly responding.

    3.1 3 Securi ty

    Installation

    To be able to install the management server and the management console on a Windows system, the installing

    user must have at least local administrative privileges on the system (It is recommended however to have

    Domain Admin privileges for that system).

  • 8/3/2019 ovowin_75_tsg

    37/91

    Copyright 2005 Hewlett-Packard

    Development Company, L.P. Page 37

    NOTE: If it is not possible to obtain Domain Admin privileges you are still able to install the OVO for

    Windows. This procedure requires a manual assignment or creation of group and user accounts at the Domain

    level prior to the installation. Please refer to the Installation Manual for details.

    The installation of the OVO for Windows management server creates two local groups HP-OVE-ADMINS and

    HP-OVE-OPERATORS (See the Runtime section below for details on these two groups).

    These groups are local to the management server. In addition, the installation procedure prompts for a Domain

    Group (the default is HP-OVE-Group) and Domain User account (the default is HP-OVE-User).

    To simplify this document this Domain Group will be referenced as its default "HP-OVE-Group", the Domain

    User as "HP-OVE-User". The actual names can be specified during the installation of the OVO for W indows

    management server and can be any valid W indows group and user name. Even existing groups can be

    selected, in this case however it has to be insured that the selected User is a member of the selected group.

    Password expiration issues

    It is recommended to disable password expiration for the HP-OVE-User. However, if password expiration is amust, OVO for W indows provides a tool to change the password in a consistent way throughout the

    installation. The HP-OVE-User should be a service account, i.e. should not be used by any real Windows user

    to log onto the system. Upon adding a node the Domain Group HP-OVE-Group is used to grant the OVO for

    Windows Management server access to a managed W indows node. The HP-OVE-Group is added to the local

    Administrators group of every managed node.

    CAUTION: Select the HP-OVE-Group and HP-OVE-User carefully. Once the selection is made, only a

    reinstallation of the OVO for W indows Management Server allows you to pick a different name.

    The OVO for W indows installation procedure will prompt for the HP-OVE-User account as well as for a

    password. If the HP-OVE-User is already available, the installation procedure will prompt for the password and

    will verify that the password is valid.

    If the HP-OVE-User is not yet available, it will be created automatically as part of the installation procedure.

    NOTE: If the HP-OVE-User is not yet created, the installing W indows user has to be a member of the group

    Domain Admins. (Windows requires Domain Admin privileges for the creation of Domain Groups and Users)

    Managing Windows/ NT/ 2000/ 2003/ XP Nodes

    Managing nodes is split into two tasks, which require different privileges:

    Adding a managed node: To add a managed node, the user running the Management Console has to have

    local Administrative privileges on any node to be managed by OVO for Windows. The easiest way to

    accomplish this is for the user to have Domain Admins rights.

  • 8/3/2019 ovowin_75_tsg

    38/91

    Copyright 2005 Hewlett-Packard

    Development Company, L.P. Page 38

    NOTE: Security restrictions may not allow the OVO Admin to have Domain Admins rights: In this case, the HP-

    OVE-Group must be added to the local Administrators group of the node to be managed manually (i.e. prior to

    adding the node).

    If the above requirements are met, a Windows/ NT managed node can be added. The process of adding the

    node will add the HP-OVE-Group to the local Administrators group on the managed node (using the toolOveConfig.exe). After pressing "Apply" or "OK" in the Node Configuration Editor, this tool is run

    automatically. If the tool succeeds either the HP-OVE-Group is added successfully at the managed node, or this

    group is already in place.

    If the tool fails, the Console running W indows user does not have enough privileges to add this group to the

    Administrator group. In this case, either add the HP-OVE-Group manually using a W indows user with higher

    privileges (e.g. a Domain Admin) or re-launch the OVO for Windows Console as this user and re-add the

    managed node.

    The OveConfig.exe application is also available via Tools at

    Tools->Administration->Windows Node Security Setup.

    It can be run either to verify if the HP-OVE-Group is set up already, or to add this group. It may be rerun any

    time for verification purpose.

    To be able to manage a node after the node is added to the management server, it is sufficient for the user

    running the Console to be an OVO Admin or OVO Op. Administrative privileges in the Windows context are

    not required for both the OVO Admin and the OVO Op. Other tasks such as deploying and removing policies

    from the managed node only require that the logged in user be an OVO Admin or an operator that is granted

    administrative privileges. The task of starting a tool only requires the user to be an OVO Admin or OVO Op.

    Note, it is also possible to preinstall the agent. The management server is then capable to synchronizing itspolicy inventory with whats out there on Windows (and UNIX) managed nodes.

    A separate white paper is available that describes management of workgroups and non-trusted domains.

    Managing UNIX Nodes (HP-UX, Solaris)

    Adding a UNIX managed node: Unlike adding a W indows managed node, adding a UNIX managed node

    only requires the user running the management console to be OVO Admin. After the node has been added as

    a managed node, the agent software installation has to be performed manually on each UNIX node.

    To install the agent software on the UNIX managed node, a root (uid=0) user privilege is required. The UNIX

    agent software installation process is always manual and cannot be initiated from the management console.

    The installation process creates a regular (non uid=0) user opc_op and a group opcgrp on the UNIX node to

    be managed.

    Note: If NIS or NIS+ is used on the UNIX managed node, it is recommended to setup an opc_op user and

    opcgrp group prior to installing the agent software.

  • 8/3/2019 ovowin_75_tsg

    39/91

    Copyright 2005 Hewlett-Packard

    Development Company, L.P. Page 39

    After installation, the agent will run in the context of the root (uid=0) user.

    NOTE: For security sensitive environments, it is possible for the agent processes to run as non-root. Please refer

    to the installation instructions in the OVO for Unix Reference Guide for details.

    3.14 Commun ication

    The MMC console communicates with the management server via WMI / DCOM. The web console uses http

    as communication means. The agents communicate via DCE RPCs with the management server. In order to

    support firewall environments, OVO for Windows 7.20 added a feature called

    DCE RPC communication without endpoint mapper with the following benefits:

    Reduce number of required open firewall ports

    Hide DCE communication, prevent DCE lookup: It must not be possible to query available endpoints

    from outside

    The endpoint mapper / portmapper / rpcd / port 135 must not be needed for agent / servercommunication

    Endpoint mapper can be stopped on agent if no other apps need it. Still needed on OVO for

    W indows management server

    Firewall filter can be implemented easier because of fixed ports and eliminated DCE lookup

    Applies to messages including annotations, actions and action responses, tools, policy and

    instrumentation deployment, agent control (start/ stop/ status), HBP

    W hite paper available since May 2003

    Communication Setup

    RPC server selects the port it is listenting to from the opcinfo file (agent) or registry key (management

    server)

    RPC server