Hotspot Overviews
Martin Moghadam
Oliver Villumsen
Yüksel Aydemir (Unlicensed)
Joachim Hansen
The hotspot overviews (ANALYZE > Overview > HotSpots > ...) are global top lists of clients, servers and networks, ordered by highest resource consumption or worst network performance.
With the hotspots you can view top-x lists (such as a top 10 list), or you can view lists that only contain values that exceed a specific threshold value. To limit your results, you can filter data on computer group as well as time.
If you select that a hotspot should cover a long period of time, PerformanceGuard may have to go through very large amounts of data in order to show the results. That's why it can sometimes take a few moments before you see the results of your search.
Learn more about selected hotspots in the following:
Client Resources
With this hotspot you can identify computers that have a high local resource consumption based average CPU usage, memory consumption, startup and login times, disk utilization, etc.
Disk-related values, such as Disk Space Usage %, concern the system disk on the selected computers.
You can either generate a top-x list (for example a top 10 list), or you can generate a list of computers that exceed a certain threshold (in which case you should select the Threshold check box).
In the search results, click computer names to view detailed information.
When search results are very large, the results table has a search field. Use the field to quickly filter your results.
Process Resources
With this hotspot you can find the most resource-consuming processes running on computers based on:
Average CPU usage percentage
Several instances of a process may run simultaneously. The average is shown per instance of a process.
If you want to easily view the average CPU usage of several instances of a process that run simultaneously, use the Computer Process Chart and select the Summarize by process name feature. That feature joins all instances of the process into one, and that lets you view the CPU usage of all of them combined.
- Average memory consumption
Average number of disk I/Os
What's disk I/O?It's short for physical disk input/output operations. While you read from a physical disk, and when you write to it, the processor must wait. Reading from a physical disk is much slower than reading from RAM. Consequently, a lot of disk I/O means that performance is slower than necessary. You can typically improve performance by optimizing the disk hardware, but that can be very complicated. A more popular approach is therefore to optimize the services that use the most disk I/O, so that they cache as much data as possible in the RAM. The Process Resources hotspot helps you find those services.
- Total number of consumed CPU seconds
Average page faults
What's a page fault?Page faults are a pretty complicated concept, but essentially page faults occur when the requested information is not where the application and Virtual Memory Manager expects it to be in the current virtual memory. When that's the case, the information must be retrieved from somewhere else, possibly from the paging file. While that happens, the thread that got the page fault is put into a wait state while the operating system finds the requested information, and that slows performance down. For more detailed information, go to technet.microsoft.com and search for page faults.
Average context switches
What's a context switch?A context switch is the process of storing and restoring the state of a process so that execution can be resumed from the same point at a later time. Simply put, if a computer has a lot of context switches, it needs to remember a lot of states, and that can affect its performance negatively.
In the search results, click a process value to view a list of all computers that run the process in question and the resource consumption on each computer.
IP data process information is not collected on computers that run Windows XP.
Client Traffic, Process Traffic, Subnet Traffic, Server Traffic
Use the Client Traffic hotspot (ANALYZE > Overview > HotSpots > Client Traffic) when you want to find the groups of computers that have the worst performance.
Use the Process Traffic hotspot (ANALYZE > Overview > HotSpots > Process Traffic) when you want to find processes that contribute to bad performance, for example the processes that generate the largest amount of traffic, or the processes that are responsible for most requests.
Use the Subnet Traffic hotspot (ANALYZE > Overview > HotSpots > Subnet Traffic) when you want to find the subnets that experience the worst performance.
Use the Server Traffic hotspot (ANALYZE > Overview > HotSpots > Server Traffic) when you want to find the servers that have the worst performance when they communicate with specific groups of computers.
You can filter the results by selecting specific computer groups, servers, ports and processes.
When you select servers for a traffic hotspot, you can only select between monitored servers.
You can generate a top-x list (for example a top 10 list). Alternatively, you can select the Threshold check box and specify the required threshold value (in which case the results will only contain values that exceed the threshold).
In most search results you can click items to view further details.
PerformanceGuard looks at the ratio between failed requests and all requests in percent. In other words: (all requests − failed requests) / all requests × 100
That depends on the type of work that you do in your organization, but you can often follow our rules of thumb.
Startup Time, Startup Processes
The Startup Time hotspot is ideal when you want to view a list of the longest startup times within a group of computers.
In the Startup Time hotspot's search results you can click computer names to view detailed information.
The Startup Processes hotspot deals with averages: It's ideal when you want to view information about the average startup time, average CPU time or average number of disk I/O operations within a group of computers.
It's short for physical disk input/output operations. While you read from a physical disk, and when you write to it, the processor must wait. Reading from a physical disk is much slower than reading from RAM. Consequently, a lot of disk I/O means that performance is slower than necessary. You can typically improve performance by optimizing the disk hardware, but that can be very complicated. A more popular approach is therefore to optimize the services that use the most disk I/O, so that they cache as much data as possible in the RAM. The Process Resources hotspot helps you find those services.
Login Time, Login Processes
The Login Time hotspot is ideal when you want to view a list of the longest login times within a group of computers.
In the Login Time hotspot's search results you can click computer names to view detailed information.
The Login Processes hotspot shows average values for processes that run during login. This way you can get an indication of whether a particular process contributes to long login times, uses a lot of CPU, or causes a lot of disk I/O operations during login.
It's short for physical disk input/output operations. While you read from a physical disk, and when you write to it, the processor must wait. Reading from a physical disk is much slower than reading from RAM. Consequently, a lot of disk I/O means that performance is slower than necessary. You can typically improve performance by optimizing the disk hardware, but that can be very complicated. A more popular approach is therefore to optimize the services that use the most disk I/O, so that they cache as much data as possible in the RAM. The Process Resources hotspot helps you find those services.
If you find a process that you want to investigate further, you can learn more about how to investigate individual processes in Monitor Network Traffic and Resource Usage per Process.
Because the Login Processes hotspot shows average values, it's a good idea to look at the number of samples in the table's third column. If there's only a few samples for a process, a single high value among those few samples could make the average look alarmingly high even though it may not necessarily be a problem in the long run.
Search this documentation
On this page
In this section