Follow BigDATAwire:
February 27, 2013

Cutting Through the Static with Model-based App Launching

Isaac Lopez

There are very limited solutions today for launching and managing apps in the cloud, says startup, Appcara, but their model-based approach can help address this next layer up the complexity chain, enabling organizations to deploy projects faster and with more flexibility than traditional approaches.

“Infrastructure clouds worry about servers and operating systems, but with a big ecommerce system with ten servers or a Hadoop cluster with tens of servers, that’s complex,” said Paul Speciale, Appcara CMO, during an interview at the Strata Conference this week. “You have to go in and instantiate each individual server, configure them, make sure they know each other – to do that server by server, it’s a nightmare.”

“Most vendors use a server template,” said Appcara founder and CEO, John Yung. “You install once, you put it into a template, and you can deploy that template many times – but that template is static. The moment you change, you need another template.”

Automating deployment using simple static server templates is no longer the most efficient way of getting your big data project up and running in the cloud, says Appcara. Instead, they suggest a model-driven approach for workloads that they’ve implemented into their Appstack platform for accelerating apps into the cloud.

“Once we capture [the information] in our data model, users can come to our marketplace with just a list of apps, click subscribe, a few more buttons, and then deploy,” says Yung. “Ten minutes later it’s operational, and because we don’t depend on templates, we can allow users to keep adding new components and continue to operate holistically.”  Appcara offers 65 apps prepackaged in various categories in their marketplace, with big data being a category. Hadoop clusters and Hive currently round out the big data category, but that’s just the beginning says Speciale. 

While there are already virtual “plug-and-play” template deployments available for Hadoop, Appcara’s assertion is that these solutions don’t scale well for evolving application environments. Yung claims that as enterprises evolve they need app deployment and management solutions that are more flexible and effective. Appcara claims their approach takes it to the next level.

“Hadoop is kind of the quintessential distributive example for us,” says Speciale. “Say a workload has 20 servers in it – click, it’s deployed. And now I want to add 5 more servers – click, click, click, it’s deployed. This is the sort of ongoing lifecycle management that we would bring to [the big data] world.”

 

Related Articles:

Red Hat Releases Storage Plug-in to the Apache Wild

Fusion ioScale Accelerates Checkouts in the Cloud

IBM Helps Move Enterprise Applications to Cloud

 

BigDATAwire