See exactly what’s happening: fast and in-depth performance analysis
ktracer enables performance analysis of processes and systems to detect performance bottlenecks, pinpoint issues and discover opportunities to improve performance.
ktracer is introduced and integrated with Caliper 5.0, which significantly improves the capability of performance analysis on HP-UX, across both the application and the kernel. This capability provides the user with an overall performance view―tracking performance bottlenecks and issues throughout the stack.
High reliability: ktracer functionality has been thoroughly tested for quality and performance internally within HP and with custom customer engagements. It is ready for deployment on all development and production systems.
Broader and deeper performance analysis: With the introduction of ktracer, HP Caliper further strengthens the functionality it provides in performance analysis covering both the application and kernel performance.
Ease of use and low performance overheads:
ktracer provides easy and flexible scripting to provide
the user with a broad choice of trace points that can be enabled.
ktracer can easily be invoked through a simple Caliper command
line option and also provides simple do-it-all options
Power and flexibility:
ktracer contains more than 40,000 customisable trace points
ktracer can be used for system-wide performance analysis or
process-specific performance tracking
Rich report generation:
ktracer augments its use by providing rich reports full of
performance data that provide easy-to-recognise performance trouble
ktracer reports locks, system calls, and variables symbolically
making it easier for you to identify programme attributes
Reports are sortable and selectable for easy navigation
How it works
Diagnose functional problems:
ktracer can easily help you recognise invalid parameters to functions
that may be the cause of performance issues with your application.
ktracer also helps to detect the failure of the kernel to release resources it may have acquired, such as locks. The timely release of these resources helps application performance and scalability.
Pinpoint issues and avoid risk:
ktracer helps to identify interrupts which are off. These interrupts may be valuable for programme execution and the disabling or enabling of these interrupts may lead to lethargic programme performance.
ktracer helps highlight error routines that are called regularly. The reporting of these routines helps the user focus their attention on it to determine its possible significance. The timely resolution of these errors may avoid potential risks to performance and quality of the application
Diagnose performance problems:
ktracer can help troubleshoot an unresponsive CPU. The
processor may be waiting for an I/O to finish during a critical
operation, in which case you would want to reschedule the I/O operation
or avoid it if possible.
ktracer also reports on thread switching. If the thread switches during a critical region, it may affect the performance adversely. You could mitigate this problem by making appropriate changes to avoid the thread switch.
Intense lock contention leads to scalability issues for the application. ktracer discovers and reports high lock contentions within the kernel.
Discover performance opportunities:
ktracer helps quantify the frequency and timing of greater than 40,000 kernel items such as reads, writes, faults, frees, connects, enqueues, and system calls. It also helps identify potential path length issues.
At a glance
ktracer is available for HP-UX 11i v3 Update 3 and later releases running on HP Integrity servers.