X
Home & Office

Yahoo spins cloud web around competitors

Company is creating a cloud computing ecosystem based on its open source Hadoop framework and hopes to lure rivals into Yahoo's universe with free, "battle-tested" software, exec reveals.
Written by Kevin Kwang, Contributor

Yahoo is looking to give away its cloud computing technology to anyone who wants it via its Hadoop initiative in order to establish its software as the foundation and standards for the "emerging cloud revolution", according to a company executive.

Speaking to ZDNet Asia in an interview Monday, Raghu Ramakrishnan, chief scientist of audience and cloud computing research at Yahoo, noted that the Internet company has not been involved in "many cloud conversations" today as these revolve around "low-level cloud services", but it is still very much in the game.

Elaborating, he said that many of today's cloud services such as those provided by Amazon Web Services and Salesforce.com, are basically infrastructure products with "baked-in" apps that can be used straight out from the box.

Yahoo, however, has chosen to take a different cloud approach, Ramakrishnan stated. The company has made a business decision to focus on monetizing its cloud computing know-how through its consumer-facing products such as its search engine and news portal, instead of offering similar cloud offerings touted in the market today, he explained.

However, this does not mean Yahoo will not change its mind over selling packaged cloud services in the future, he acknowledged.

With regard to continually refining its consumer products, Yahoo is working with developers through its open source initiative Hadoop, which is the company's analytics and data warehousing framework, to constantly improve its compute capabilities, said the executive. The Hadoop framework was made available to the open source community last June and, since then, Yahoo has contributed about 70 percent of all code written, he added.

For instance, Yahoo's content optimization knowledge engine (Coke), is an algorithm programmed using Hadoop's technology to determine the top stories featured on the company's Web site on a daily basis, Ramakrishnan said.

"By including factors such as what news articles users read previously, their search preferences and the type of applications they enjoy using, Coke will be able to determine which story should be 'pushed' to readers when they first enter the site," he explained.

This has helped the company experience an increase of 160 percent in click-through rates for its targeted news stories compared to having a "one-size-fits-all" approach, the executive said.

In terms of other cloud vendors making use of its open source framework, Ramakrishnan said not many companies would "turn down the chance to use software worth tens of millions of dollars that are battle-tested at Yahoo scale" and resell these products to their customers.

Social-networking sites such as Facebook and LinkedIn, for example, make use of Hadoop to store and analyze the data stored in their servers, according to him. Amazon launched its Hadoop-based Elastic MapReduce service, which is targeted at businesses, researchers and analysts involved in number-intensive data crunching projects, in April 2009.

However, the company has no regrets in availing its technology to anyone who wants it, Ramakrishnan said.

"We're not looking at making money through the selling of [software] licenses. We gain from these companies [that use our software] when they buy into our ecosystem," he said. "Effectively, we are helping ourselves by contributing to what we hope will become the software foundation and standards for the emerging cloud revolution."

That said, he admitted a desire to see more recognition being given to the company for the "under-the-hood" work it is doing to drive cloud computing technologies.

Developing a semantic search engine
As for projects that are in Yahoo's pipeline, the scientist revealed that one of its "secret sauces" that the company is in the midst of brewing is in the area of using semantics to refine end-users' search processes.

He said that his team is currently looking to draw links between users' search patterns as well as content they had previously posted in their social-networking sites and the language they use in those posts, which offer contextual clues. These links, in turn, could eventually provide the search operator a better understanding of people's needs in terms of search capabilities, Ramakrishnan added.

"We're not the only ones looking to unravel the mystery of understanding the world in semantic terms. Google is also working on this," he said. "However, this is a great challenge that will not be solved in the next 5 to 10 years' time."

Ramakrishnan also pointed out that Yahoo's search partnership deal with Microsoft has been "misrepresented" by the media.

"We believe that certain search functions are pretty standard today and, by partnering with Microsoft, we are passing these services on to them so that we can focus on higher-end search services [such as its semantics project]," he explained.

Yahoo in August announced it had completed the transition of its main search results in the United States and Canada over to Microsoft's search engine, Bing. The latter now powers Yahoo's Web, image and video search for both the desktop and mobile platforms.

Editorial standards