Case studies

A major Pacific Rim online gaming enterprise

 

A major Pacific Rim online gaming enterprise learns how to optimize its resource utilization to lower bandwidth demand while not compromising performance    
A major Pacific Rim online gaming enterprise learns how to optimize its resource utilization to lower bandwidth demand while not compromising performance
   

Business impact

Reduced the bandwidth demand and improved stability on the Web site without having to upgrade the IT infrastructure or revamp the software applications.

Issue

The online gaming enterprise operates a high volume Web site for publishing information and processing transactions. It needed to expand its IT resources to support those applications with an eye toward future growth without incurring the costs of additional servers, network devices and increased maintenance.

Executive summary

IBM Research and IBM Global Business Services worked in partnership with the client to analyze its Web site traffic and bandwidth demands. The IBM team devised a solution to help redistribute demand and boost Web site performance. The result was a reduction in page weights that facilitated faster page-loading to help improve customer experience. The client reports no load-related crashes since the IBM recommendations were implemented.

What IBM did

The online gaming enterprise Web site routinely handles large numbers of service subscribers and concurrent users, which caused occasional network slowdowns. IBM assembled a team of experts to analyze the network's workload, both at the network and the application layers. The analysis was designed to identify: the pages that consumed the most bandwidth; the bandwidth utilization pattern; and the relationship between the number of online users and bandwidth demand.

The client used two sets of Web servers, one dedicated to providing real-time information and the other to handling transactions. Because the end user pages refreshed frequently, the information-providing servers carried the heavier workload, despite the client's attempts to ameliorate the situation through the addition of cache servers and load balancing switches. The results of the bandwidth demand analysis indicated that the transaction server bandwidth demand was only a fifth that of the information providing server. The heavy demand on the information providing servers was attributed, in large part, to the frequent update of several XML, Javascript and HTML files.

The IBM team used the analysis to devise a multi-faceted solution to help reduce the bandwidth demand on the information servers. To shrink the size of the XML files, the IBM team recommended replacing the XML name tags with shorter ones (for instance, replacing <team_info> with <t> or <ti>) as well as reducing the depth of the XML tree by including some XML sub-elements in the father node. To shrink the size of the HTML and javascript files, the client was able to use common reduction techniques, such as filtering out white space before the files reached the server. In addition, the client's Web site's page update mechanism was changed to process only incremental updates, i.e. replacing only those portions of the information that changed. Finally, GZip compression was used to compress the HTML, Javascript and CSS files. The result was a reduction in page weights, faster Web site performance and a dramatic decrease in load-related server crashes.

Capabilities applied

IBM researchers and IBM Global Business Services consultants used cutting edge network analytical tools and technology expertise to devise low-cost ways to help optimize the client's e-business processes through resource utilization optimization, while paving the way for growth using existing hardware and applications.

Related case studies

A major communications company

IBM Research uses cutting edge technologies and methods to assess networks and network applications for this communications company.


More on Issue

Network performance management

The Network Performance Management specialty uses a combination of performance monitoring techniques to help businesses meet performance, availability and resiliency goals.


More on research

Bandwidth problems in high speed networks

IBM researchers offer solutions to four common bandwidth problems.