Jennifer Case Studies

Credit Card Company
Telephone Company
Web Home Shopping Company
Web Event Ticket Sales Company
Insurance Company

Company Type: Credit Card Company

Company Description:
Subsidiary of a large finance holding company. Main business is consumer credit and debit cards. 13.1 Million members, Assets: $18B, Capital $626.8M

Web Server Configuration:
IBM Websphere, Z/OS

Why web site performance is important to the on line credit card business:
If the credit card transaction fails or transaction are delayed (customer is waiting for the system to respond), the probability of customers abandoning the credit card transaction rises significantly; customer chooses another method of payment or cancels the transaction all together. This results in loss of customer and their business. Thus, the credit card company IT staff must monitor business critical system in real-time in order to proactively deal with application performance problems as they are detected.

Problems faced by IT team:
Needs to have an effective error determination/resolution process to identify and correct specific causes for "failed transaction", "delayed transaction time" and "intermittent service failure". Finding the root-cause of the problem was not an easy task. The IBM WebSphere on z/OS that performs most of business critical tasks was extremely difficult to analyze, earning the nickname 'BlackBox' by the IT staff. Determining the root-cause of the performance problem became very time consuming. The performance problems also often lead to finger-pointing and hurt feelings between the SI managers and the system developers since the root-cause of the problems are often unknown and difficult to determine. The company needed a solution to help the SI team discover and document the problem so the system developers could resolve it.

What features should an Application Performance Monitoring system provide?
Most business critical applications perform transactions that are on CICS on WebSphere z/OS. The APM solution of choice must be able to clearly monitor the transaction which starts from web request and end in external transaction through CICS in WebSphere z/OS. The APM solution must fully support IBM WebSphere running on z/OS, support real-time monitoring for the other operating systems and web application server types. It also must provide robust root-cause analysis and resolution features for performance problems and errors.

The Benchmark test proved Jennifer was the best solution:
To find the best APM solution, the team developed a BMT (Benchmark Test) The IT team examined several APM (Application Performance Management) solutions including JENNIFER. In the test, most solutions were not installed successfully, or even when installed, had limited capabilities. After BMT, the company selected the JENNIFER because it was easy to install, received the highest score in BMT and had the best reputation amongst the local APM solutions. The IT team leader said that "when the performance problem is detected most of the other APM solutions simply analyze the problem using log data or system commands alone. However, JENNIFER allows user to detect and analyze the performance error by providing the detailed profiling data related to the error."

Benefits using Jennifer:
After adopting JENNIFER, the company has reduced the number of system errors by 70% and reduced the time spent resolving the errors by 80%. The IT team was able to establish an effective transaction monitoring process in real-time for the IBM WebSphere for z/OS – this prevented customer and revenue loss by minimizing system downtime and service delay - Maximize system availability by establishing the performance problem management process - Offer pleasant online service experience to the customer, increasing customer satisfaction - Provide a standard for the system performance management when integrating new systems.

Back to Top

Company Type: Telephone Company

Company Description:
The company is a Mobile Telephone service provider with 50% of the Korean market. The company is a market leader in state of the art mobile 3rd generation mobile phone service and on-line shopping.

Web Server Configuration:
OS: SUN, Solaris
WAS: Weblogic
DB: Oracle
Applied System Environment: 85 Servers / 220 CPUs/ Mobile Phone Shopping / Services

Why web site performance is important to a telephone company:
In the rapidly changing mobile phone service business, the company is adding new features and services rapidly. Customers expect each new service to respond quickly as they move between functions and applications. Delays or errors badly damage customer satisfaction resulting in lost customers, lost revenue and lost mobile phone market share.

Problems faced by IT team:
The company needed a tool to help them manage their increasing IT infrastructure, especially in WAS area. As the number of servers increased, the IT team was faced with more frequent system performance problems; the amount of work they had to do to constantly administer and maintain the servers seem like an impossible task. In order to manage multiple different systems environments that provide variety of different services (home shopping, customer account management, website, mobile phone services etc...), the company needed an APM solution that allow the administrator to monitor all systems in a single, elegant view that is comprehensive in providing vital information about the system environment but also simple and intuitive enough even for a WAS novice to use and understand. They also needed a solution with the ability to monitor services and catch system errors for 24x7 x 365 in real time bases. Finally, they wanted a monitoring solution which the monitoring screen can be accessed easily from anywhere with convenience.

What features of Jennifer were of greatest value?
Jennifer was able to collect and utilize data from 36 different systems in a single view through Jennifer's comprehensive and intuitive Dashboard screen. With Jennifer's Dashboard, system administrators achieved resolution of existing application performance problems as well as prevent the future performance problems, and enhance system efficiency by tuning bad site code and inefficient DB queries. Jennifer also provided administrators with email and SMS alerts whenever an error is detected in the system so that administrators are informed of any problem ASAP. Finally, system administrators were able to access Jennifer monitoring screen from any computer with a web browser and web connection, allowing administrators to use Jennifer from virtually anywhere.

Benefits using Jennifer to the IT team
Before Jennifer, development team and operations team had a difficult time working together when faced with a performance problem. Because there was no data to help them point their attention in the right direction, the development team and operations team did not know what was causing the performance problem and who should fix it. With Jennifer, both teams were able to collect relevant data that helped clearly identify the root-cause of the problem, preventing conflict and promoting interdepartmental team work. Before Jennifer, the IT team spent 2 weeks per each system performing diagnosis and running performance tests. After Jennifer was implemented, all performance data was available to the IT staff in real-time, eliminating the needs to run additional performance tests and diagnostics, saving valuable time and IT resource.

Back to Top

Company Type: Home Shopping

Company Description:
The company is the third largest on-line retailer in the world, trailing QVC and HSN. Total Sales for 2008 exceed $500M. The business includes a 24-hour TV home shopping through the cable and satellite broadcasting, the largest Internet shopping mall in Korea, the largest on-line retailer in Korea and a new on-line shopping service in China.

Web Server Configuration:
OS: SunOS 5.9, Solaris
WAS: IBM WebSphere 5.1
DB: Oracle 9i
Applied System Environment: 9 Servers / 50 CPUs /Internet shopping mall

Why site performance is important to an on-line shopping company:
Service response time and website performance plays an important role in internet shopping. Delayed service response time or website performance problem translates into loss revenue and customer satisfaction. According to an internet market study, if an internet store fails to provide service to customers, approximately 35% of the customer will not return to that store. This means that when ever a internet store is unable to provide service to customers due to delayed response time or performance problem, it is losses not only the revenue from the present transaction but future business as well. Therefore, immediate resolution of any delayed response time or performance problem is needed for an internet store and mall to be successful. There can be dramatic increases in the number of visitors (example: sales events or gift giving holidays such as Christmas, Valentines Day, etc...) which can stress the network. When the incoming service request exceeds internet store capacity, the incoming request must be redirected to prevent system overload or system crash. The overflow must be redirected to a proper error page instead of long delay in service response time.

What was required by the IT department to support the network?
The IT team must be able to monitor performance problems, response time, system load, and system resource in real time. It would be a big advantage to see as many of these elements in a single graphical presentation. Administrators need to be alerted of any errors or performance problems immediately via SMS messaging as they signal the company is losing new on-line orders. Management expects the IT team to identify and resolve the performance issue as soon as they occur, 24 x 7 x 360. In rare cases of overload, service requests must divert visitors to an error message which will have a more positive impact than an unresponsive screen. Every lost or disrupted transaction is an immediate loss of revenue for the company.

How did Jennifer benefit the company?
Using Jennifer, the IT staff was able to setup a monitoring system that will collect relevant monitoring data in real time and ensure system stability. This data is presented in a graphical summary screen where it is easy to identify problems by simply drilling down to get more specific data. The collected monitoring data was delivered to the site developers for continual repair and improvement of the sites. Also, PLC (Peak Load Control) played a significant role in maintaining system stability during the busy seasons when the incoming service request exceeds the website capacity (diverts PLC requests to an error page). An added benefit is the data collected from Jennifer is used for marketing and business purposes as well as server maintenance purposes.

Back to Top

Company Type: On-Line Ticket Sales and Distribution Company

Company Description:
The company is the only online ticketing service in Korea that provides services for the entire line of domestic events such as concerts, movies, sporting events, and tourist attractions. The company provides services from the Box offices at the event as well as other channels such as internet, call center, mobile phone and other communication infrastructure. The 10 year old company has 70% of the domestic ticket sales in Korea, more than $10M in annual revenue and over 1300 visitors daily.

Web Server Configuration:
OS: Linux
WAS: 40 Resin Servers
DB: Oracle 9i

Why site performance is important to an on-line ticket company:
Customers who purchase tickets on line have a very short attention span. If the site is not responsive, they quickly switch to another site. The site must be able to handle huge peak loads when special events are offered and thousand of visitors try to get the "good seats".

What problems did the company have before Jennifer?
An internet site ranking and consulting firm had reported that the site visitor count for the company had increased 30~50% during fourth week of November as the holiday season approached. That indicated the 60 servers now in use may become overloaded. The company web page's average service response time had doubled since the dramatic increase in visitor count during November. Visitors were seeing a blank screen when trying to load a webpage, causing frustration and dissatisfaction. They needed a tool that can be used in real-time during operation hours, collect and store service statistical data for performance problem analysis, and connect to backend system to retrieve monitoring data and resolve performance problem as well.

How was Jennifer able to help the company?
By applying PLC (Peak Load Control) to Ticket Link Site, the company obtained the means to prevent system crashes in case of sudden flux in site visitor counts by redirecting customers to the proper webpage. By redirecting users to an alternative webpage at peak periods (rather than let them stare at a blank page), the dissatisfaction and frustration felt by the site visitors was minimized. Also, by identifying and resolving errors and exception in the system with Jennifer, the company was able to minimize service error counts experienced by customer and improved site transaction speed by tuning application and database queries.

Back to Top

Company Type: Insurance Company

Company Description:
The company is the third largest private marine, fire, automobile insurer in Korea with business also in Japan, China and the USA. Total Assets for 2008 exceed $5B. Total employees exceed 3,000 with over 16,000 agents.

Web Server Configuration:
OS: AIX 5.3
WAS: WebSphere 6.1
DB: Oracle 9i

Why site performance is important to an insurance company:
With tens of thousands of agents, the company must provide easy to use and responsive on-line tools so the agents and customers using the website can quickly process new insurance policies. If the application for a new policy takes too long because of poor response time or errors, the company may lose the customer.

What did the IT team need to manage the website?
Once the IT team completed development of a new application, they needed an Application Performance Monitoring tool to track how it performed with the company agents in the field and web visitors. It was a key management goal to keep agent and web site user satisfaction high. The burden for this fell on the IT team. They needed to collect and manage various data in real-time to locate problems that may require modification of application code. Also required was monitoring of the backend system's performance and service status. The goal was to locate the root-cause of errors as quickly as possible so the application developers could resolve the issue.

How did Jennifer help the IT team keep the insurance agents happy?
Jennifer filtered application errors/exceptions for each multiple task and provided vital information to resolve such errors. Therefore, error frequencies are significantly decreased and service speed is improved thanks to creating an appropriate tuning after diagnosing performance problem. By quickly locating and repairing errors, the field agents, web site visitors and management were happy with the performance of the website and the on-line tools used by the remote agents.

Back to Top