Interop/Software 2008 addresses distributed performance challenges inherent to PLM

It's impossible to miss the fact that large global enterprises as well as midsize manufacturing companies have been centralizing their IT resources for several years now. They do so not only for the projected economies of scale, but to ease compliance management across multiple fronts.

With systems scattered the world over, “Companies didn't know where their servers were, and they didn't know what was on them,” says Alan Saldich, a VP with Riverbed Technology.

For CAD and product life-cycle management (PLM) vendors, however, the flip side of enhanced control through centralization is that “distributed performance management has become their number-one problem,” says Saldich. It can take hours to send CAD files from, for example, a design center in California to one in Israel, and the productivity loss involved is only exacerbated by significant time-zone differences.

That doesn't mean PLM vendors are hoping for a quick pendulum swing back to the “good old days” of decentralization.

In the past—and undoubtedly still—use of distributed, local data stores caused significant revision control challenges for design groups. The poster-child example is Airbus, and the huge losses it incurred when poorly communicated wiring harness changes made it impossible to fit fuselage sections together for the initial instances of its flagship A380 jet.

At this year's Interop/Software 2008 conference and expo, it was clear that at least several IT infrastructure providers—including Riverbed, Akamai, and FastSoft—have “felt the pain” of manufacturers and are gearing their solutions to address the problem. Yet the companies seem to be taking decidedly different approaches. 

Several infrastructure providers have noted the need in today’s manufacturing enterprises for solutions that allow expeditious sharing of 3D models. FastSoft’s Aria solution, when installed on the host network, accelerates TCP traffic for all desired applications to all destinations.

Akamai's EdgePlatform is one of the world's largest distributed computing platforms. Its network of more than 25,000 servers in 69 countries relies on applied mathematics and algorithms to relieve Internet congestion and vulnerability problems. The servers reside within approximately 900 of the world's networks, where they monitor the Internet to gather information about traffic, congestion, and trouble spots. This “intelligence” is used to optimize routes and replicate data dynamically to deliver content quickly and securely.

Riverbed thinks more enterprises will turn to wide-area data services to improve performance of PLM and other applications across a wide area network (WAN). Riverbed solutions accelerate digital content delivery by taking an integrated approach to application performance across the WAN. Specifically, an application-independent foundation that improves the performance of all applications running over TCP is combined with an application-specific set of modules that address chatty application protocols.

Riverbed's Steelhead appliances enhance application performance via these type activities:

• Data streamlining optimizes WAN traffic by removing redundant data and prioritizing traffic;
• Transport streamlining improves TCP behavior; and
• Application streamlining reduces application protocol inefficiencies and enables disconnected operations.

FastSoft says its hardware-based approach allows companies to distribute content up to 30 times faster than otherwise. Unlike some other acceleration solutions, FastSoft's E Series requires no hardware or software on the receiving end. Benefits of the core technology, called FastTCP, include the following:

• Native TCP enhancement is fully standards-compliant;
• It is exceptionally resilient to packet loss and latency; and
• No software installation or kernel modifications are required.

The focus of these infrastructure providers on the needs of PLM users reinforces the growing importance of 3D models in an increasing range of enterprise business processes.