Users optimize capacity with grids, Part 2

Chalon Mullins wanted to stop building extra capacity just for spikes in traffic on the Charles Schwab network. He also needed to move the company’s infrastructure off expensive servers to less-expensive boxes while enabling Web services across multiple locations – all without making the network vulnerable to downtime.

“Downtime gets Charles Schwab on the front page of The Wall Street Journal,” Mullins, technical director of infrastructure strategy and architecture at Charles Schwab, told Comdex attendees last month.

Mullins, along with other speakers, detailed how he worked with the Globus Alliance and its Globus Toolkit, an open-source implementation of the Open Grid Services Infrastructure (OGSI) specification, to build grid-enabled applications. Grid computing proponents say the technology lets multiple computing resources share processing power and resources to enable enterprise IT managers to get more out of the servers, storage and systems they have in their networks.

For his part, Mullins looked into grids to help Charles Schwab spend less to do the same or more with services running on its infrastructure.

According to his presentation at the show, Mullins in 1999 thought of grid computing more like “science fiction,” but in 2000 he realized grid computing fit in with Charles Schwab’s plans for adopting a service-oriented architecture to support Web services. With two data centers, 14 mainframes and about 700 servers, Mullins said a grid was an option that would tap into the “white space,” or unused capacity, across multiple resources. Mullins and his team got to work “capturing unused computing cycles” and applying a grid application to enable processing across the two data centers and to share resources among the 700 or so servers.

“We used to build excess capacity into our infrastructure to ensure availability during peak load times,” Mullins said. “When we started to see commercial support for grids and the Globus toolkit, we realized it wasn’t just an academic endeavor.”

Mark Wirt, chief technology officer at Butterfly.net, an online gaming service provider, said his company looked into grid computing because of the worldwide network of distributed applications it supports. Yet right now, the company is only using grid applications to support back-end process management and load balancing that doesn’t touch Butterfly.net gamers.

“The game players don’t touch grid services, but when they do, the grid application will be transparent to our users,” Wirt said. Butterfly.net is using its grid application to more closely tie together the company’s game service providers, which are located in cities around the globe.

“It’s still a nascent industry, but grid applications can become a way to provide resources to multiple locations fluidly,” Wirt said.