DTrace (Dynamic Trace) of Solaris™ 10 Operating System function

Solaris™10 : [ Solaris Containers | Solaris ZFS | Predictive Self-healing | DTrace ]

If performance deteriorates or other server system problems occur Dynamic Trace (DTrace) helps you resolve the situation. DTrace lets you monitor and understand the operational state, detailed system behavior and system problems, on your server.

Winner of the Gold award in Wall Street Journal's Technological Innovation Award in 2006, DTrace was evaluated top in break-through technology against more than 600 applications. Its minimal workload and around 40 thousand tracing points imbedded in the Solaris™ 10 Operating System, provide powerful trace filtering functions. These allow you to monitor system behavior in detail - obtaining just the right information anytime you need.

Before DTrace, resolving performance problems was a time-consuming and complicated task. In addition traditional system analysis tools could in themselves affect system performance. This meant that system administrators often had insufficient opportunity to obtain useful analysis information. As a result they had to presume the cause and then confirm their presumptions by trial and error and their own experience.

Traditional system analysis tools were also primarily aimed at the debugging of single applications. This often made them inadequate at investigating the behavior of an entire server. This in turn further frustrated the efforts of the administrator.

For instance, kernel information tools such as process status lists or system call traces, although able to provide important clues, don't directly lead to resolution due to the coarseness of information provided. With intermittent faults, memory snapshot, which is often the investigator's course of last resort, is hard to use effectively so takes a significant time to reach a conclusion.

Designed as an analysis tool for enterprise servers, DTrace enables extraction of the relevant information and works alongside operational systems on the server without interruption and with minimum performance degradation.

Characteristics

DTrace, lets you obtain both kernel information and library information enabling you to fully investigate from both OS and application viewpoints.

The 40 thousand trace points in the OS mean you don't need to write your own debug programs. They also let you obtain relevant records close to the fault site. This ensures faster cause detection and swift resolution of application and system bottlenecks. Plus because you can filter out irrelevant trace points you can maintain good system performance. Now you can analyze your system at your site, without any need to build a clone system for troubleshooting.