Chat now with support
Chat with Support

Spotlight on DB2 6.10 - User Guide

Spotlight on IBM DB2 LUW (Linux, Unix, and Windows)
New in This Release Getting started with Spotlight on IBM DB2 LUW Desktop features specific to Spotlight on IBM DB2 LUW Spotlight on IBM DB2 LUW drilldowns
About Spotlight on IBM DB2 LUW drilldowns Buffer Pool Analysis drilldown Client Application Analysis drilldown Database Analysis drilldown Database Manager Summary drilldown Diagnostic Log drilldown FCM Analysis drilldown Tablespace Analysis drilldown Top SQL drilldown Operating System drilldown Workload Management Analysis drilldown
Spotlight on IBM DB2 LUW alarms Spotlight on IBM DB2 LUW Options Tuning SQL statements in Spotlight on IBM DB2 LUW
Spotlight Basics
Spotlight Connections Monitor Spotlight Connections Alarms Charts, Grids And Home Page Components View | Options Troubleshooting
Spotlight History Spotlight on Windows
Connect to Windows Systems Background Information Home Page Alarms Drilldowns View | Options Troubleshooting
Spotlight on Unix About us Third-party contributions Copyright

System Panel

 

Displays information about the operating system of the machine you are currently connected to.

OS — Operating system — The version name of the Windows operating system that is installed on the target machine.

Ver — Version — The full version number of the Windows operating system installed on the machine.

SP — Service pack — The name of the service pack installed on the machine (if any).

 

Related Topics

Network Panel

 

The Network panel shows the total number of connected users, and the rate at which packets are being sent from and received by the system:

Windows Networking Users — Shows the number of clients connected to this system.

This does not show users connected to other applications that may be running on this machine (for example, Microsoft Exchange or SQL server), only the users that have established a Microsoft Networking connection to the system.

Open Sessions — The current number of Windows Networking sessions open on this machine.

A session is defined as a user connection to a Windows Networking resource. For example, a connection to the ADMIN$ and C$ shares on a server by a single users is counted as two sessions.

Theoretical Bandwidth Limit — Shows the level of network traffic graphed against a "theoretical" maximum bandwidth for the network card specified. If there are multiple network cards, use Windows Network Card Display to select the one whose data you want to display.

Alarms can be raised in this panel for all network cards in the machine, not only the card displayed.

Note: Because of the methods involved in sending data, the actual maximum on an Ethernet network usually is approximately 50% of the theoretical bandwidth. The actual bandwidth can equal the theoretical bandwidth only when using full duplex Ethernet.

Token-passing network topologies (including TokenRing and FDDI) enable a theoretical maximum bandwidth much closer to the actual.

Selected Network Card — The name of the network interface card (NIC) selected for diagnosis. If there are multiple network cards, use Windows Network Card Display to select the one whose data you want to display.

 

NBT KB Received — Represents the rate at which NetBios TCP/IP data is being received by the system from Windows Networking clients.

NBT KB Sent — Represents the rate at which NetBios TCP/IP data is being sent from the system to Windows Networking clients.

 

Total KB Received — Represents the rate at which network data is received by the system from other network systems.

Total KB Sent — Represents the rate at which network data is being sent from the system to other network systems.

 

Related Topics

Event Log Panel

 

The Event Logs panel provides a link to the Event Log drilldown.

  • It alerts you to items that have reached alarm status as per your configuration of Event Log alarms.
  • The button icon shows the number of Event Log alarms that have been raised but not yet cleared. You can acknowledge and clear alarms via the Event Log drilldown.

Note: This button is enabled on Enable Event Logs selected. Windows Event Log Metrics

 

Related Topics

Windows Event Log Metrics

CPU Panel

 

The CPU panel shows processor and load information for the Windows system:

CPUs — The number of processors installed on the target system.

Speed — The speed of the processor(s) installed on the target system.

Type — The type of processor(s) installed on the target system (Intel Pentium III, for example).

Up Time — Shows how long it has been since Windows was started on the machine being diagnosed.

Total CPU Usage — This spinner represents the total amount of CPU being used on the machine being diagnosed. It includes CPU consumed by all windows processes. This component spins faster as CPU demands increase.

Processor Queue Length — Shows the average number of threads (program execution units) that are waiting to run on each processor. A sustained average greater than three can indicate processor congestion.

Processes — Shows the number of applications and services that are currently active on the system.

Threads — Shows the number of threads currently active in all processes. A thread is a running portion of the owning process. Every process has at least one thread.

 

Page Outs per second — Shows the number of memory pages per second that are being written to disk from physical memory, in order to free up physical memory. The memory pages are checked first to see if they have been modified - if no modification has taken place, they are discarded.

Page Faults per second — This is a measure of the number of page faults per second on the system. This value includes soft faults and hard faults. A page fault occurs when a process requires code or data that is not in its space in physical memory. A soft fault is when a memory page is not in the physical memory of the process, but resides in another segment of physical memory. A hard fault occurs when a memory page is not in the physical memory of the process, and has to be brought from disk.

 

Cache Writes — This flow shows the rate at which cache pages are written to disk from memory, in order to satisfy a write through request (an application request to not write information to cache, but directly to disk) or a cache flush (a regular writing of data from the write-cache to disk).

Cache Reads — Shows the rate at which cache pages are read by applications from memory.

 

 

Related Topics

Related Documents

The document was helpful.

Select Rating

I easily found the information I needed.

Select Rating