by Rob Barry and Jack Vaughan
To get a view of how cloud computing may progress, one may look at the course of data grids and distributed caching. Boutique companies such as Appistry, GigaSpace, and DataSynapse have plied the parallel computing trade for a good while, and cloud computing seems a very natural next step.
By submitting your personal information, you agree that TechTarget and its partners may contact you regarding relevant content, products and special offers.
Lesser known due to its roots in the narrower Microsoft software market is ScaleOut Software, which last month unveiled a Management Pack for its ScaleOut State Server that includes an object browser and parallel backup/restore capability intended to help architects and developers view, manage, and back up objects stored in its distributed cache. In recent releases, ScaleOut has expanded beyond .NET to support Java distributed applications as well.
ScaleOut-style grid infrastructures may eventually enable so-called hybrid cloud environments. The company’s CEO, William Bain, has founded several start-ups, including Valence Research. There, he developed a distributed Web load-balancing system that was acquired by Microsoft and is now called Network Load Balancing in Windows Server.
While Bain has focused on developing what he calls a “distributed data grid,” hybrid cloud environments factor heavily into ScaleOut’s strategy.
“What makes the grid so useful in the cloud [is that] distributed data grids naturally have an elastic capability,” said Bain. “It’s a perfect fit. A grid will grow with an app and allow it to have high performance without the bottleneck of having to access a server or a [Binary Large Object] store.”
ScaleOut’s recent release takes a distributed grid and gives users the ability to manage, sort and mirror any and all cached data. This should prove helpful as cloud architectures migrate data and applications from single and multi-location grids to public clouds. Allowing IT teams to analyze and report on objects inside the distributed cache of StateServer is an important step toward mainstream use.
Amway, a company in direct sales and marketing, is currently using ScaleOut to build an international e-commerce application. The company is migrating off a 60-server .NET system tied into a WebSphere application server tier. In the existing environment, the cache was duplicated on each Web server.
“My division was created to build a global Web site,” said Mark Andrews, supervisor of commerce applications in the Global E-Business division of Amway. “We’re in 80 different markets and territories but each one has its own IT operation and its own Web site. The focus of this project was to consolidate that so that eventually everyone’s working from a common framework. We’re using ScaleOut to limit the amount of times we have to talk to those external services.”
Andrews knew Amway would need a distributed cache for the project. He said the company looked into Microsoft Velocity, but the product didn’t have the managing and monitoring capabilities he was looking for. He said ScaleOut was relatively simple to set up and had a number of the management features he was used to having to build in-house.
Bain feels strongly that the qualities that separate a middle-of-the-road distributed caching system from a top quality one involve the software’s ability to interpret what’s going on inside the cache. Open-source distributed caching systems like Memcached enable extremely large datasets to flow through Web servers at high spee, but the lack of transparency in the system provides a bottleneck to performance configurability, suggests Bain, who said the Management Pack can remedy such ills.
With so much interest moving to hybrid clouds, distributed caching vendors seem to be putting their products in position to be bridges to the cloud.