This section can help you to understand the characteristics of operating system agents and agentless monitors, as well as the differences between them.
IBM® Tivoli® Monitoring provides operating system (OS) agents that monitor the availability and performance of the computers in your monitoring environment. An example of an OS agent is the monitoring agent for Windows, which can monitor Windows XP, Windows 2003, and Windows 2008 operating systems. These full-function OS agents must reside on the same computers they are monitoring.
- API
- Function
- SNMP
- The Simple Network Management Protocol is a TCP/IP transport protocol for exchanging network management data and controlling the monitoring and operation of network nodes in a TCP/IP environment.
- CIM
- The Common Information Model is an XML-based standard for defining device and application characteristics so system administrators and management programs can monitor and control them using the same set of tools, regardless of their differing architectures. CIM provides a more comprehensive toolkit for such management functions than the Simple Network Management Protocol.
- WMI
- Microsoft's Windows Management Instrumentation API provides a toolkit for managing devices and applications in a network of Windows-based computers. WMI provides data about the status of local or remote computer systems as well as the tools for controlling them. WMI is included with the Windows XP and Windows Server 2003 and 2008 operating systems.
These APIs are supported by the Agent Builder, which enables you to build custom agentless monitoring solutions that are separate from the agentless monitors available on the Tivoli Monitoring installation media and that provide additional function.
Since an agentless monitor is a standard Tivoli Monitoring agent, it collects data and distributes it to a Tivoli Enterprise Monitoring Server and then on to a Tivoli Enterprise Portal Server. It also takes advantage of the various features of the IBM Tivoli Monitoring product, such as Tivoli Enterprise Portal workspace views, situations, remote deployment of the agentless monitors, policies, and so on. Detailed information can be found in the user's guide for each agentless monitor; see Table 1.
Agentless monitoring does not provide the kind of deep-dive information your site may need for its core business servers; however, it does allow a small set of centralized servers to supervise the health of the operating nodes in your environment. There are five types of agentless monitors that cover the Windows, AIX®, Linux, HP-UX, and Solaris environments.
The agentless monitors are multi-instance agents. After installing or deploying an agentless monitor on a machine, additional instances can be created via configuration. Each instance can communicate with up to 100 remote nodes.
Each type of agentless monitor can run on additional platforms beyond the type of platform it monitors. For example, the agentless monitor for Windows (which monitors only Windows operating systems) can run on any of the supported platforms: Windows, AIX, Solaris, HP-UX, Linux.
Specific operating system releases that a particular agentless monitor can monitor are detailed in Table 1. Check the user's guide for each agentless monitor regarding platform-specific requirements for the operating systems that agentless monitors can run with.
A computer that has one or more agentless monitors running on it is referred to as an agentless monitoring server. Each server node can support up to 10 active agentless monitor instances, in any combination of agentless monitor types; for example, 2 AIX, 2 HP-UX, 2 Linux, 2 Solaris, 2 Windows; or 4 Windows, 3 AIX, 3 Linux; or 5 Windows, 5 Solaris; or 10 HP-UX. Each instance can communicate with up to 100 remote nodes, which means a single agentless monitoring server can support as many as 1000 monitored systems (10 instances * 100 remote nodes per instance). By adding more server nodes, the number of monitored nodes increases into the thousands.
Agentless technology provides lightweight OS monitoring that targets key metrics along with basic situations meant to satisfy simple monitoring needs. Agentless monitoring provides speedy implementation and minimum agent deployment, including the deployment of updates; however, the need to poll the monitored node to retrieve its monitoring data increases network traffic, and real-time data availability is impacted both by the network delay and the reliance on polling. In addition, the implementation of Take Action commands for command and control is more powerful with the full-function agents than for agentless technology.
- Logical and physical disk utilization.
- Network utilization
- Virtual and physical memory
- System-level information
- Aggregate processor utilization
- Process availability
- Disk utilization
- Memory utilization
- CPU utilization
- Network utilization
| Agentless monitor | Product code | Data collectors supported | Operating system releases monitored |
|---|---|---|---|
| Agentless Monitoring for Windows OS | R2 |
|
Monitors the following Windows releases:
Note: IA64 machines running Windows are
not supported.
|
| Agentless Monitoring for AIX OS | R3 | SNMP V1, V2c, V3 | Monitors the following AIX releases:
|
| Agentless Monitoring for Linux OS | R4 | SNMP V1, V2c, V3 | Monitors the following Linux releases running on xSeries®, pSeries, and zSeries® machines:
|
| Agentless Monitoring for HP-UX OS | R5 | SNMP V1, V2c, V3 | Monitors the following HP-UX releases:
|
| Agentless Monitoring for Solaris OS | R6 |
|
Monitors the following Solaris releases:
|
| 1 To use one of the native Windows data collectors (WMI, PerfMon, the event log), the agentless monitoring server must run under Windows. | |||
IBM recommends that you deploy a full-feature operating system agent to each agentless monitoring server to watch the CPU, memory, and network consumption of the agentless monitors themselves.