Hadoop, Exalead and The Future of Computing
For the past two years, the promise of cloud computing has brought an intense interest in technologies that support large-scale, distributed computing models.
This excitement? Definitely warranted.
BusinessWeek technology writer Stephen Baker covered parallel computing technology Hadoop in December of 2007 in his story “The Two Flavors of Google”. Baker wrote “As more businesses and researchers shift complex data operations to clusters of computers known as clouds, the software that orchestrates that teamwork becomes increasingly vital.”
Hadoop and similar technologies are an important first phase of a computing evolution.
At Exalead, we developed our own computing framework for our web search engine (found today at Exalead.com/search). We call our parallel computing framework DSort. Exalead’s DSort technology, like Hadoop, distributes a computation over a large number of machines.
This scalability is appreciated by a number of clients. For example, France’s blogging platform, Skyblog, which ranks 16th on comScore’s list of most popular web sites in the world, manages 500 million blog entries and handles more than 20 million search requests daily.
The real excitement for Exalead, however, is the idea that we can combine this scalable architecture (in our case DSort) with a level of intelligence and information access functionality that most people haven’t seen before.
This is a major step towards the next big phase of computing.
As analysts from The 451 recently wrote, Exalead appears to be at a crossover point between search and BI. “The challenge with this idea is of course that for the past 30 years or so relational databases have been where the ‘important stuff’ has been stored and the multi-billion BI market grew on top of that as a way to access it.
Database administrators rule(d) the roost as far as information management goes. Meanwhile enterprise search got relegated to a side room where it was all about finding documents and getting pages and pages of results returned to you.
What..Exalead and a few other companies are moving towards is a convergence of the two; call it database offloading, unified information access; unified information intelligence or something similar.”
This means companies are creating mission critical business applications – apps that were impossible a few years ago — on our solution. These applications can scale to extraordinary levels at a fraction of the cost – and often times can accomplish more than traditional IT technologies.
Web-based Architecture Moves Into the Enterprise
One example: GEFCO, a €3.5 billion company which ranks among Europe’s leading transport and logistic firms, adopted Exalead CloudView for use in their new application: “Track and Trace.” The application offers powerful search functionality and up-to-the-minute information from an extremely large data set.
“Before we installed Exalead CloudView it could take a day to get the results of such a CPU-intensive query, by which time the information was out of date,” said Guillaume Rabier, Manager of Studies and Projects at GEFCO. “Now we get these answers almost instantly.”
“With Exalead we are using database off-loading to intelligently rationalize the use of our resource systems. By maintaining a separate index, fewer of the users’ queries run against the database itself. This in turn reduces the load on the database servers which generates results faster and produces greater IT efficiency.”
Hadoop is worth all of the excitement.
We think the direction that companies like Exalead are taking will be worth even more.









March 13th, 2009 at 1:49 am
Sorry, to spoil the party, but if applications would be developed the right way – based on a business architecture – and content is only produced as it should be – in the business context — all that hype about Googling your apps away could be ignored. Yes, search is a quick-fix for the mess that enterprise systems are in, but if it turns out to be like Google with 7 millionhits for a search .. why bother? Let’s do it right!