MapR's Apache Hadoop distribution, an open source Big Data platform, has broken performance barriers and enjoyed widespread adoption by major enterprises, the company says.
Is Hadoop secure, fast and reliable enough to meet the production Big Data needs of enterprises around the world? Some vendors, such as RainStor, seem to think Hadoop may still be a little immature. But MapR made its alternate viewpoint clear this week at the Hadoop Summit in San Jose, where it announced significant performance benchmarks for its Hadoop distribution and flouted its widespread adoption by big-name companies around the world.
The benchmarks were the result of testing using the ioMemory platform from Fusion, which provided a high-performance storage system for calculating read speeds of a MapR Hadoop cluster. The company said that read-intensive applications in the test were 25 times faster than on traditional Hadoop implementations.
While MapR didn't specify how much of the performance increase might have been due to the storage optimizations of the Fusion platform, it emphasized that the benchmarks point to important new possibilities for Hadoop by demonstrating a way of overcoming limitations associated with traditional disk storage. "This record is significant because it eliminates one of the barriers to HBase adoption in high-performance environments," said Bill Bonin, vice president of Business Development at MapR.
At the same time that it highlighted the impressive performance of its Hadoop distribution in conjunction with non-traditional storage, MapR also enumerated some of its biggest-name customers to pitch its product as "the de-facto standard Hadoop distribution for industry leaders." In particular, it mentioned the following companies:
Organizations that have switched from another Hadoop distribution to MapR include Ancestry.com, the world's largest online resource for family history; Quicken Loans, America's largest online mortgage lender; Live Nation, the world's largest ticket seller and promoter of live entertainment; Recruit Technologies, the number one (sic) provider of recruiting services in the world; Cision, the market leader in PR software and media tools and the world's leading source of media contacts and information; ReturnPath, the worldwide leader in email intelligence; Solutionary, the leading pure-play managed security service provider; and The Rubicon Project, the leading technology company automating the buying and selling of advertising globally.
MapR's purporting to have become the de-facto standard of Hadoop is a big claim, and it may be a little early to speak of total market dominance when other Hadoop distributors, Hortonworks included, are eagerly competing for their slice of the pie. Still, the companies MapR counts among its customer base demonstrate its penetration of a significant portion of this growing market.
They also suggest that enterprises are increasingly seeing Hadoop as ready for prime time. Security and reliability concerns may remain somewhat of a hindrance, and value-added Hadoop distributions like MapR's may consequently enjoy more momentum than more generic implementations. Still, Hadoop is faster and more popular than ever, and will only become more so as it matures further.