Case Studies


Case Study # 1: Global Semiconductor Manufacturer.  Locating unknown CPU consumption

It started as a request to demo onTune from the system administrator, a member of the company IT staff, who had a serious performance problem with the management servers of a new factory line. Concurrently, they were also using an Application Performance Management (APM) solution to try to analyze and resolve their problem.

Using onTune a short time after installation, the system administrator discovered a strange CPU-consuming pattern in their application. The development team was informed of it , and eventually found the cause of the performance problem starting from this clue.

The customer’s feedback was that onTune provided much more informative and practical results. onTune provided much more useful information than the APMs, which in this case were almost useless. Furthermore, the installation process for APMs were very difficult, cumbersome, and took a long time in comparison to onTune installation. onTune installation was accomplished within 30 minutes for 16 servers.

Actually, while the management and maintenance of the company’s servers were outsourced, yet by using onTune, the IT staff were able to analyze their servers in detail, much more so than the outsourcing company’s professionals. The outsourcing company felt challenged by this experience to review onTune in comparison to their own in-house monitoring solution. The result was a decision by the outsourcing company to replace their own monitoring solution with onTune for their entire server population, which numbered about 2,000 servers.

Following this decision, the outsourcing company also decided to extend onTune use to the customer’s overseas factories (in China and USA). They felt their ability to analyze these systems would be enhanced and strengthened even though there is the added cost of onTune.

P.S.: Even though servers in factories are operating 24 hours a day, IT staff and resources are unable to keep tabs on these servers for 24 hours a day. If something does occur during their absence, it’s practically impossible to analyze which process(es) caused the problem. Available monitoring tools in general are only able to provide simple performance data sampled over a long monitoring interval. onTune is able to provide exact (correct, appropriate, and useful) performance data to analyze any performance incident even for short intervals, like 2 seconds.


Case Study # 2: Stock Exchange. Monitoring system performance in seconds intervals

A few years ago, the Stock Exchange replaced their old system with a new generation system. At the time, they decided to use onTune for their new system, even though they already had an installed (general) monitoring system. The existing monitoring system was insufficient and impractical.

Their system’s main function is stock trading, and a normal stock trade transaction should complete within one second. (Although a one-second stock trade transaction completion is not necessary, it’s imperative for stock trade companies to provide this service as rapidly as they can, as this increases their market competitiveness.) They therefore required their IT staff to be able monitor and analyze the system in intervals of seconds, at the very least. This requirement was even more critical during the starting time and ending time of stock trading.

For the customer in this case, there was no better solution than onTune. onTune was chosen and implemented, and is serving them well in almost all areas and aspects of monitoring, analysis, and reporting.


Case Study # 3: Leading Telecommunications Company – Monitoring very large server configurations.

This customer upgraded their service offering a few years ago, migrating from unix servers to linux servers, and using the oracle DBMS system.

 

The service’s main task was to calculate the length of a phone call, and the corresponding fee to charge to the phone user. To provide this service, each linux server was a huge system with 80 CPU cores, and 500 GB memory. Furthermore, a short response time is necessary as the charges information would be browsed via smart phone frequently. (Smart phone users are always looking for, and expecting quick response, more so because they are using small mobile devices, and outdoors.)

It has always been difficult and a challenge to monitor or manage the customer’s huge servers. Moreover, the customer always needed more detailed, more accurate performance data for use in analysis and reporting. They had created and were using their own monitoring software, but it was insufficient and failed to meet their requirements. Finally, after a thorough review they decided on onTune for their use. Once they decided, and installed and started using onTune, they have been very satisfied, so much so that onTune has become a standard for new servers. So now, every new server deployed includes an onTune license for that server.


Case Study # 4: Retail Chain Monitoring and alarming anomalous CPU pool conditions in virtualized IBM servers.

The IT staff at this well-known retail chain was overwhelmed by spurious alarms from their performance monitor. Their brick and mortar and online business runs on about 100 LPARs with half for test and development, and half for production. They wanted a system performance monitor that does everything they needed, but none of the fancy stuff. This was particularly true of their need to monitor the resource usage in CPU pool of virtualized IBM servers. Using the onTune Event Manager, the customer is able to set anomalous condition limits where they are alarmed for these conditions if the CPU pool dedicated to Oracle processes and programs.

The reliability and accuracy of the performance data collect by onTune provided the basis for the event alerts and alarming of anomalous conditions. To provide this service, the menu-driven onTune Event viewer allows the user to define and set events based upon system resource utilization; processes running on servers; text pattern in log files; file system usage; scanning of service ports; and so on. Event notification and alarming are flexible and easily defined.

The end result is the IT staff and users can trust the event alarms they receive, and are freed from the distraction of false anomalous conditions.

 


onTune SPA – the most capable System Performance Monitor an Analyzer for IBM virtualized server environments !

“onTune SPA has all the functionality we need, and none of the fancy stuff we don’t want…” Retail Customer

“I can’t understand how I did it before without onTune SPA…” Semiconductor Manufacturer Customer