Gorodenkoff - stock.adobe.com
Hadoop distribution provider MapR is prepared to shut down operations as it looks for a buyer or backer on the heels of "extremely poor" financial results for its fiscal quarter that ended in April.
These intentions were disclosed in a letter to employees and a notice MapR filed on May 13 with California's Employment Development Department, saying the company could possibly shutter its headquarter offices in Santa Clara and terminate 122 employees there if needed funding wasn't obtained by June 14.
MapR's website says the company has more than 450 employees worldwide, but California law only requires advance notifications of possible layoffs in that state.
A private firm, MapR even just last year had been seen as a potential initial public offering (IPO) candidate and had recently reorganized its sales group. Once commonly named among big data high-fliers, the company was a pioneer in bringing the Hadoop data processing framework to enterprises. It lists Audi, Cisco, UnitedHealth and others among its customers.
From its outset, MapR took a distinct tack in that it forged its own data file format, one that was POSIX-compliant and which was not directly aligned with the Hadoop Distributed File System forwarded by others, notably Cloudera and Hortonworks.
MapR also developed other proprietary technologies, including its own database and event streaming system, instead of relying on open source options.
Its name aside, MapR, like the other big data players, increasingly augmented Hadoop's original MapReduce engine, incorporating Apache Spark and other Hadoop-related components into a fairly broad product line.
Its present troubles and the uncertainty about MapR's future put users in jeopardy. "It is hard to speculate, but there are a good number of customers with large MapR deployments that will be keen to find out," said Doug Henschen, analyst at Constellation Research.
A general trend toward cloud-based big data implementations has challenged MapR and other players associated one way or the other with the Hadoop movement. The company's inability to quickly gain further financing or close such a deal now seems to encumber its once promising outlook.
What has happened recently, Henschen indicated, is big data has evolved, with cloud services capturing most of the growth and on-premises deployments of Hadoop lagging as the slowest part of the market.
Henschen cautioned that MapR isn't a Hadoop vendor, but a "commercial software vendor focused on combining performance and scale." Still, he added, "it was up to MapR management to avoid ending up in a financially vulnerable position."
Hadoop and similar data processing formats helped drive big data implementations at leading web companies, like Google and Facebook. MapR stridently specialized in helping mainstream enterprises move to Hadoop.
But across the Hadoop ecosystem, prototypes and pilots have tended to be the norm, rather than full on-premises production implementations.
"Unfortunately, it seems MapR is being tarred with the same brush that's declaring Hadoop to be dead," Henschen said. "But the data is not going anywhere soon. The need to manage big data is by no means dead."
MapR's flagship Data Platform has displayed useful enterprise features for data protection, management and resilience, according to Mike Matchett, analyst and founder of the Small World Big Data consultancy.
"To some extent, they have had to spend a lot of time building products that were simply me-too or slightly improved versions of software that was either pure open source or simply competitive to Cloudera," he said.
Doug HenschenAnalyst, Constellation Research
Matchett highlighted attention MapR has placed on tools such as its Drill SQL-on-Hadoop implementation, which competes with Impala from Cloudera, Presto available from Starburst Data and others, and a host of other SQL-on-Hadoop alternatives. Differentiation in this market is challenging, he said.
"MapR has been in a difficult position for a long time. Technical advantages they have gained in performance, for example, they have not been long-lasting, and their claims of being more enterprise-ready haven't stood up so well as other companies' products matured," said Curt Monash, president of Monash Research, an IT consultancy firm.
Like others in the big data space, MapR has expanded into diverse product categories, said Monash, who cited analytical database management systems as a major example.
"Hadoop companies have had considerable success there, but, for financial markets, it hasn't lived up to the hype," he said.
Roadmap is rocky
Following IPOs and a general diminution in Hadoop enthusiasm, coupled with increased competition from AWS and other cloud platform providers, Cloudera and Hortonworks merged in January. This led many to wonder if a merger or acquisition might be in the offing for MapR, which was often cited as the third player in a Hadoop distro triumvirate led by Cloudera and Hortonworks.
"Clearly, the powerhouse combination of Cloudera and Hortonworks increased the competitive positioning challenge for MapR," Matchett said. "Yes, they went from No. 3 to No. 2 overnight, but the competitive gap must have gotten larger."
In a letter to employees, MapR CEO and Chairman John Schroeder said the company had been near to obtaining needed financing, but that an unnamed supporter backed out when the company saw reduced sales for the quarter ending April 30. The search for financial support continues.