Benchmarking CRM Software
RadView Benchmarks to Test Blue Martini Performance
24 September 2001
RadView Software and Blue Martini Software have completed a eCRM benchmark using a fully-functioning customer web site.
The benchmark measured the Blue Martini eCRM application suite maintained sub-second response time - from page request to complete page load - while scaling up to 47,000 concurrent users, and sub-two second response time with user loads significantly exceeding these levels.
The performance benchmark tested a retail customer site built on the Blue Martini application suite, with a catalog of over one million stock keeping units (SKUs). The benchmark simulated realistic user scenarios and think-times, and returned real pages in response to user requests and actions.
"Generating a volume of 47,000 concurrent users to test an application far exceeds the current average test size for an e-business environment,
which typically simulates no more than 10,000 concurrent users" said Billie Shea, Director of Research at Newport Group Inc, a leading performance testing analyst organization.
Dan Murphy, executive vice president of US Operations at RadView explained:
"The Blue Martini Software benchmark on the Sun platform represents the most significant load test undertaken to date in the eCRM space. The breadth and depth of the application, running in a real-world environment, consistently performed and scaled, and we are pleased that our WebLOAD software was selected to drive this significant test."
"We have experienced conversion rates that are 3 times the industry average, and have retention rates above 50%. Our rapid growth requires the highly scalable and reliable eCRM applications and infrastructure that Blue Martini provides," said Rowan Gormley, CEO at Virgin Wines, the UK's premier online wine retailer.
Benchmark Test Methodology
The benchmark test was designed to capture the performance metrics of a high volume, rich content, real world site running on Blue Martini applications and was performed using the following parameters:
-- User model - the benchmark used scripted user behavior modeled with a realistic distribution of time spent browsing, updating, buying, searching, waiting, etc. In between actions, the virtual user waits an amount of time ranging from 4 seconds to 10.5 minutes based on an industry standard "think time" distribution curve
-- Store size - the test used a real-web store with one million SKUs in inventory, more than 5 times larger than a typical large web store
-- Store design - real-world page design with multiple images per page
-- Response criteria - in the benchmark, WebLOAD measured response time from the moment the user clicks a link or button to the time when a full page is returned. The goal was to achieve sub-second responses across a wide range of load scenarios
-- Hardware configuration - the benchmark tested realistic hardware configurations that included load balancers, web servers, application servers, and database servers. Tested horizontal scalability configurations with multiple instances of Blue Martini applications on multiple application servers
www.radview.com
www.bluemartini.com.
Delicious
Digg
Reddit
Magnoliacom
Newsvine
Furl
Google
Yahoo
Technorati
Comments
Post new comment