Two major players in the application acceleration market turned things up a notch when they announced upgrades...
to their products at Interop Las Vegas this week.
Citrix Systems announced its new Citrix NetScaler 12000 system, an upgrade to its applications acceleration product, which targets high-end Web applications handling millions of simultaneous users conducting hundreds of thousands of transactions per second. The company claims the product processes application transactions up to 400% faster than its competitors.
Rather than focusing solely on network-layer performance and packet processing, the Citrix NetScaler 12000 is designed to deliver application-specific performance gains for complex and demanding Web applications, the company said. The NetScaler 12000 accelerates application transactions, with more than 275,000 Web (HTTP) requests per second and 28,000 secure (SSL-based) transactions per second.
"There's not a person out there who doesn't think about ways to improve the performance of the application," said Zeus Kerravala, vice president of infrastructure and security research at Yankee Group in Boston. "The [technology] approach is less important than the application specificity. It's important to say you improve Oracle or SAP 'this way' versus just generically solving all application problems."
For its part, Crescendo announced multi-tier application acceleration technology for Web-based applications. Application Layer Processing (ALP), an addition to its Maestro family of products, is technology that accelerates application flows across logical application tiers. The product optimizes all application tiers in order to address end-to-end data center performance bottlenecks, regardless of location. The product incorporates Maestro's short life transaction (SLT) technology, which addresses front-tier issues.
According to Crescendo, ALP -- using definitions created by the Crescendo Rule Engine (CRE) -- is capable of recognizing for which tier each application request is destined. ALP also understands that different requests impose different processing "weights" on the application, with "heavier" requests taking longer to process than "lighter" ones. Request weights can either be manually configured or adaptively learned by ALP. In addition, ALP recognizes that each tier in the application has an upper processing limit in terms of simultaneous requests. When a tier reaches maximum capacity, ALP's Admission Control mechanism queues requests within Maestro, sending them to the application only when processing capacity is available.
"What is crucial is how you improve application transactions and what are the metrics tied to applications transactions," said Hooman Beheshti, vice president of technology at Crescendo. "Traditionally, this is not how people have thought about application acceleration, but as we talk to more buyers and analysts we realize that this is how [users] think. It used to be networking guys making buying decisions, but now it's the application guy making the decision to buy it. The SAP guy versus the networking guy in charge of Cisco gear."