App Acceleration Not a Pipe Dream

Networking vendors ramp up acceleration efforts as newcomers come to the fore with new ideas.

 By Sean Michael Kerner
Page of   |  Back to Page 1
Print Article

Networks with the fastest servers or the biggest bandwidth pipes don't necessarily deliver any given application the fastest anymore. New application acceleration appliances and services hit the market recently, with vendors shooting right at faster application delivery.

Citrix is pushing its fastest-ever appliance while Juniper is pronouncing its vision and introducing a new management tool. Cisco jumped the gun a bit and announced its Application Control Engine (ACE) last month.

Then there is the number of smaller vendors vying for their own pieces, including Crescendo, which claims that it is the first company to address application multi-tier acceleration.

Crescendo's ALP (Application Layer Processing) targets application acceleration inside the data center by intelligently accelerating the application flow between all tiers.

Hooman Beheshti, vice president of technology at Crescendo, explained that most acceleration approaches physically sit in front of the front-most tier of the application, usually the Web tier.

As such, enterprises have been concentrating on performance bottlenecks from the edge of the network outbound. Content compression, TCP optimization, caching, and load balancing are examples of functionality that helps deliver content once it's generated.

But there is a need for more than just a Web tier of acceleration.

Beheshti said that applications are multi-tiered entities with application logic and database components often sitting behind the Web tier. Content is generated through processing across every application tier.

But while existing application acceleration solutions have addressed the processing time a bit by performing offload functionality for the Web tier, Beheshti said they don't address performance bottlenecks in the back-end processing tiers.

"ALP is the first technology to actually control request admission, queue requests to prevent application overload, reschedule heavy requests and intelligently reschedule them using definitions created by the Crescendo Rule Engine (CRE)," Beheshti said.

ALP is not without its barriers though, most of which by Beheshti's own admission reside within the deploying company.

He said that in order to optimally accelerate the application, there needs to be a certain level of communication between the network group and the applications group.

"This will take communication with the people writing the applications," Beheshti said. "Historically, there has not been perfect communications between those corporate groups. We are seeing those barriers come down steadily."

Juniper Networks has its own aggressive application acceleration vision.

A key part of that vision involves not just accelerating application delivery but actually measuring and monitoring the system used to perform the acceleration.

Juniper's WX Central Management System (WX CMS) has been updated to version 5.2, offering the promise of improved functionality to help enable enterprises to roll out and optimize their application acceleration efforts.

There are a number of reasons why application acceleration has recently become a hot topic.

Mike Banic, director of product marketing at Juniper Networks, said that part of the increased focus on application acceleration is the enterprise emphasis around business process and the need to increase efficiencies in those processes.

According to Banic, enterprise CIOs are beginning to realize that it's the performance of an application that prevents efficiency, which in turn is driving the decision to deploy solutions.

"There is also a lot broader knowledge now that these kinds of solutions are perfect if you're centralizing servers and consolidating data centers," Banic said. "And because that's such a broad trend, we're seeing a great amount of uptake."

Greg Smith, director of product marketing of the application networking group at Citrix, noted that there is an increasing need on the customer side as the company rolls out more applications to a broader set of users that needs ever-increasing performance.

"I also think that you're seeing advancement in the technologies that are available to enterprise customers," Smith said. "I don't think there are any technological barriers; I think what you're seeing in the market today is that there is a pretty dramatic shift to upgrade application infrastructure."

Smith said that there are a lot of large enterprises that made investments in older-generation, layer 4 load balancers back in 2000 and 2001 that provide adequate traffic management but don't provide the ability to accelerate traffic.

Customers are looking for more than just traffic management. According to Smith, they are looking at things such as data compression, caching, TCP optimization and TCP multi-plexing.

"We are not necessarily exclusively focused on networking performance metrics like packets per second and raw system throughput; that's what differentiates us," Smith said. "We're about processing and delivery of application data, not about making the network perform."

Article courtesy of internetnews.com

This article was originally published on May 9, 2006
Get the Latest Scoop with Networking Update Newsletter