Azul Solution Eliminates Unpredictable Website Performance While Reducing Complexity and Costs

Starting off as a chain of independent retail stores at the start of the 20th century, this upscale retailer became renowned for offering the finest quality men’s and women’s fashions throughout the United States.

In the summer of 2000, the company launched its customer website, offering an enhanced, personalized, online shopping experience.

Online Presence Drives Need for Scalable, High Performance and Scalability

With a stellar reputation for providing luxury goods from a wide array of designers, the goal of the company’s website was to offer an extensive variety of in-store categories while delivering the same legendary service and style for which it was known for.

From the outset, the company’s goal was to ensure a positive customer experience on the customer website. However, the company underestimated the potential success of this sales and service channel, especially since consumers were becoming more comfortable with online purchasing.

Very soon after launching the site, the website experienced a steady increase in traffic and transaction volume. As a result, performance and response time was becoming increasingly unpredictable. In addition, due to the various vendor products installed, the complexity of the native deployment and numerous components made the environment difficult to monitor, troubleshoot, and manage upgrades.

Further, the existing server infrastructure was consistently running at over 80% CPU utilization, leaving little headroom for scalability. Even with extensive caching being added to the application, it made little difference to the overall user experience.

Pilot Program Reveals Performance Bottlenecks

These performance issues jeopardized the success of the website, a vital and expanding sales and service channel for the retailer. Moving quickly to address these concerns, the company engaged Azul Systems to design a pilot program to benchmark an Azul-based solution versus the native deployment. The original website was based on an infrastructure consisting of WebLogic, Blue Martini, Tangosol and Endeca products. The native deployment required 6 Sun v220.

The goal of the pilot was to build a proof-of-concept system, which when deployed in production, could eliminate the performance problems, while demonstrating at least the same level of redundancy as the current production environment.

The pilot was designed with two tests: a Product Array and Web Checkout application. The Product Array test performed random user navigation on the website and selects random items to view. This test applied concurrency pressure on the underlying cache layers and the database. The Web Checkout simulated gift card purchases, requiring users to sign into the application and validate address information.

The pilot was an overwhelming success, demonstrating that a single WebLogic instance on an Azul appliance could perform approximately twice the number of transactions as all nine of the existing Weblogic instances combined.

From this successful test model, a production environment was deployed, with two Azul 3310D models and a third appliance for development. Each production Azul Appliance runs two instances of WebLogic in a cluster, for added redundancy.

Azul Solution Offers Simplicity and Predictable Response Times

The retailer realized several benefits immediately from the Azul solution. First, the solution enabled the company to remove four Sun V220’s, three WebLogic instances and three Tangosol instances from the original production environment. This not only provided better performance but also offered a much simpler and more cost-effective deployment.

Secondly, end user response time improved by an average of 66%. In addition, the unpredictable performance was completely eliminated.

For the company, Azul Compute Appliances provide an unmatched software and hardware-assisted solution that enables applications to transparently tap into virtually unlimited capacity over the network and reach unprecedented levels of scalability, throughput, and performance consistency.

This breakthrough scalability is achieved with unique capabilities such as hardware-assisted Pauseless Garbage Collection, which eliminates the impact of garbage collection-related application pauses, and Optimistic Thread Concurrency, which minimizes the impact of scalability bottlenecks caused by memory lock contentions. This scalability also enables the company to contain datacenter costs while ensuring quality customer experiences, especially during unpredictable peak loading such as seasonal periods, a common concern for online retailers.