It is no secret that the big three commercial Hadoop distributors have been running into headwinds in recent years as more workloads and data have made their way into the public cloud and that these Hadoop platform providers have spent a lot of money and time to expand their stacks beyond the basic open source Hadoop.
Hadoop was a very good way of enabling enterprises to store and analyze lots of unstructured data from multiple sources, and evolved it add more structured data format and even SQL query functions across those formats, all of which is clever and useful. However, Hadoop has been under pressure as cloud service providers like Amazon Web Services, Microsoft Azure, and Google Cloud Platform have offered similar capabilities in the cloud without the hassle and cost of managing the software and infrastructure to do all that on-premises – sometimes based on Hadoop, sometimes not. The rise in adoption by organizations of multicloud and hybrid cloud strategies convinced some in the industry to question the larger value of Hadoop, even as more capabilities – like machine learning and Spark in-memory processing – was added.
Two of the major Hadoop players – Cloudera and Hortonworks – turned to each other last fall in a $5.2 billion merger in hopes of creating a stronger company and ending the financial toll their fierce competition was taking on each other. Both had gone public over the past several years – Hortonworks in 2014 and Cloudera in 2017 – to raise money. Their stock prices have fallen more recently. Now the other top Hadoop vendor, MapR Technologies, appears to be weeks away from shutting its doors unless more investors jump on board or another company buys it.
In a letter sent on May 13 to employees, California state officials, and the city of Santa Clara where MapR is headquartered, John Schroeder, the company’s founder, president, and chief executive officer, said that the company will close down operations and layoff 122 employees by June 14 unless more financing comes through or “other strategic transactions” happen.
MapR, which has remained private even though it had announced its intention to go public about four years ago, has raised more than $280 million dollars since it was launched in 2009 and at one point had a market valuation of more than $1 billion. In 2017, the company raised $56 million in funding and has almost 70 customers, including major names like Audi, Cisco Systems, Cision, Ericsson, Hewlett Packard, Novartis, SAP, and UnitedHealthcare. However, MapR over the past month has been jolted by bad news. According to Schroeder, the company saw “extremely poor results” in its first quarter, which ended April 30.
“Our Q1 results were not only disappointing, but also unexpected,” Schroeder explained. “While the reasons for the results are not entirely understood, they were at least in part due to the sudden, last-minute, and unexpected postponement of several customers’ timelines to making a purchasing decision.”
The financial results had a ripple effect. Before the numbers came in, MapR was negotiating with another company for financing its debt, which Schroeder said would have allowed the company to continue operating for the foreseeable future. The negotiations had reached the point where MapR and the other organization had entered into a time sheet and Schroeder had expected the financing to come in May or June. However, that funding source pulled out of the deal after seeing MapR’s first quarter numbers. Since then, MapR has been looking for other sources of financing and for potential buyers. Its board of directors is reviewing letters of intent from two different organizations, though it is unknown whether any deal will come through in time to avoid the closure and layoffs.
Schroeder’s letter to employees and local and state officials was released under California’s Worker Adjust and Retraining Notification (WARN) laws. Schroeder in his letter said MapR did not give the notice required by the state regarding layoffs because he believed MapR had a “realistic opportunity” to get the financing needed and avoid layoffs. The domino effect of the poor financial results and the potential investor pulling out of the deal changed that.
MapR added more than 50 new customers last year and was being used in an autonomous vehicle program. The company this year also shed some direct sales and marketing jobs as its looked to streamline operations.
At the same time, MapR has looked to grow beyond its roots in Hadoop by expanding into such areas as containers, Kubernetes and other technologies for collecting, managing and analyzing data. That has included announcing support last year for AWS’ Elastic Container Service for Kubernetes and introducing the MapR Data Fabric for Kubernetes, which offers easier access to data across clouds and on-premises deployments. In February, MapR rolled out MapR Ecosystem Pack version 6.1, which included the implementation of the Container Storage Interface (CSI) for Kubernetes. Most recently, the company made its MapR Data Platform more container-friendly by making it easier for enterprises to deploy Drill and Apache Spark applications in Kubernetes
However, all this didn’t mean that the company doesn’t see a future for Hadoop. Suzy Visvanathan, senior director of product management at MapR, outlined several scenarios for The Next Platform that showed that not all data will reside in the cloud and that many workloads would continue to run on-premises, where Hadoop can be a factor because it is relatively cheap and yet relatively malleable storage.
One scenario is that enterprises still have large numbers of legacy applications that are being restructured to become more cloud-enabled or to take advantage of microservices, containers and Kubernetes, a process that will take several years. Also, on-premises environments will continue to play a role in mulitcloud and hybrid-cloud initiatives. In addition, enterprises will see their in-house data management expertise grow as they bring aboard more data scientists, which will keep some data and applications on-premises and continue to drive the need not only for Hadoop but also containers and Kubernetes.
However, the cloud has continued to stalk MapR and other Hadoop vendors, as has the challenge of offering proprietary solutions in Hadoop platforms and in an IT world that is increasingly moving in the direction of open technologies. While MapR has embraced the open source Hadoop and Spark software, the core file system that always was the differentiating underpinning of the MapR distribution, and which eventually gave it the ability to be transformed into a database, remained closed source. This may be been a contributing factor to slower adoption of the MapR distribution – it is hard to say for sure, however.
MapR has about two weeks to determine whether it can clear its financial hurdles, and if not, it looks like it will have to shut its doors. Which would be a bad thing for innovation for the Hadoop platform.