Accelerating Web applications, Part 2

When choosing a Web application acceleration product, network managers have a long laundry list of things to consider, according to Meta Group.

In a recent white paper, the research firm discusses the many issues with Web application performance and narrowed them down to three main culprits: bandwidth congestion, server latency and network latency. With multiple products available from a slew of vendors, Meta Program Director Peter Firstbrook provides the Web application acceleration criteria network managers should consider when looking to buy a product to speed app performance.

To start, Meta says the product should accelerate applications and minimize the cost of infrastructure. That means to accelerate applications network managers shouldn’t have to buy a whole bunch of boxes to do compression and caching and so on. Adding boxes can mean additional work for network managers in terms of maintenance and support, and, depending on product architecture, more appliances could represent more points of failure.

Yet Meta also recommends products that are easy to install, configure and test – and are basically not a management hassle. Sometimes appliances do offer those qualities.

Secondly Meta says the acceleration products must be “completely transparent to the client or application,” which means they do not require application code changes or client software to work successfully. Plus, the product needs to be reusable across different end users. For example, Meta says, the product should be applicable to branch-office, remote-access, business partner and customer end users. And the product must be reusable across multiple applications and a variety of application types.

Another factor in considering a product is the type of traffic being transported. A good acceleration tool will speed encrypted and plain text equally. Plus, the product shouldn’t shift the bottleneck from the Web server to the back-end application server or database. And lastly, it should address more than one performance limitation.

Firstbrook goes on to detail the manageability requirements of the product selection. As mentioned above, the product should be easy to test, configure and install. It should also provide reporting and “operational visibility through internal software health monitoring and integration with network monitoring tools,” Meta says.

Products should also be fault-tolerant with “fail-to-wire” technology so that if and when the acceleration appliance fails it will not affect application performance or cause any downtime. In addition to that, the product should actually provide high-availability features and scale to operate in a distributed or clustered environment. Essentially, Meta says a network-based application acceleration product that performs one of multiple tasks (compression, caching, SSL offloading and so on) could represent the most versatile and appropriate option for a majority of enterprise companies.

Meta concludes: “IT organizations dealing with slow Web applications must consider the source of the problem as well as the deployment scenarios to correctly identify appropriate solutions. Network-based application acceleration products can improve the performance of challenging applications while alleviating the need for incremental bandwidth upgrades, additional servers and/or application code modification.”

The full white paper is available here (free registration required):

https://www.fineground.com/products/white_papers/