Why Google Capital Placed Its Hadoop Bet On MapR

Captured by uploader. (Photo credit: Wikipedia)

Now that MapR has received a load of cash from Google Capital, it is time to ask: What does this latest round of Hadoop funding mean? All three of the major players have gotten large amounts of money from well-known investors. What can we infer from looking at who funded whom?

To my thinking, Google Capital's investment means that the start of the race is over. There are three major players, MapR, Cloudera, and Hortonworks, with Pivotal, WANdisco, Syncsort and dozens of others selling superchargers and enhancements. But the main event will be between the major players, each of which has a different strategy.

To win, each of the players needs to frame the competition a different way. It is already clear that all three companies are getting traction. And with Hadoop and MapR funded with total funding of just short of $200 million each, and MapR funded with more than $1 billion, it is clear that all have staying power. Remember, it is likely that all three can win in the sense that the current employees and investors get a solid exit. What is more interesting is understanding how Hadoop will succeed in the enterprise, where most of the money will be spent. Here's a run down of the race as I see it.

Hortonworks: Embeddability and Open Source Purity

I always imagine the Hortonworks offices as full of monks in orange robes. Five times a day, a bell rings and everyone meditates about shining void that is the power of pure open source.

This image resonates because the Hortonworks strategy is based on two elements. First is that they are a pure open source company, that doesn't sell any proprietary software. This claim alone is enough to attract those in the enterprise who love the open source model and want to avoid lock-in of any sort. Hortonworks wins deals on this proposition alone, but it is not enough for them to win.

The real driver of Hortonworks strategy is safe embeddability. Hortonworks has partnerships with Microsoft and Teradata who have blessed the Hortonworks distribution and are incorporating it into their data warehouse ecosystems. These larger players have a massive installed base and well-developed channels that will introduce Hadoop into thousands of accounts.

Hortonworks offers to be an unthreatening guide to Hadoop for the data warehouse ecosystem giants. Hortonworks can help influence the development of the open source projects to better adapt to the data warehouse ecosystem. Hortonworks also can assert it is the least threatening of the three major Hadoop players because it will never develop proprietary capabilities to compete with the data warehouse ecosystem.

Clever sales staff from Hortonworks often use the analogy of Linux and Solaris, the two flavors of Unix, the former open source and the later originally developed as proprietary operating system before it was released as open source. The message would be, 'We are Linux. They are Solaris. Remember, open source always wins.' While this analysis has superficial bite, it fails to recognize that the vast majority of all three of the major players is made up of the Apache Hadoop core distribution.

Hortonworks is betting that any gaps in enterprise usability will be addressed by their data warehouse ecosystem partners. Cloudera to some extent, and MapR in a big way, are arguing that these gaps are far larger than Hortonworks assumes and require more serious remediation.

Cloudera: First Mover Momentum

I think of Cloudera as a Wall Street hedge fund, legitimate visionaries who trade ahead of the market. Mike Olson, founding CEO of Cloudera and now Chief Strategy Officer, is the Gordon Gekko figure, in suspenders behind a desk waving around a massive cigar. Amr Awadallah, the CTO, is the head trader, thinking of ways to get ahead.

I recently wrote about Cloudera in two articles (' Cloudera's Strategy for Conquering Big Data in the Enterprise', ' Cloudera's Cash is Not an Adequate Business Strategy ') and afterward spoke to Mike Olson. My analysis was critical of Cloudera's strategy which I summarized as follows: Cloudera is seeking to fill the gap between enterprise needs and what the Apache Hadoop offers by creating a small amount of proprietary software to enhance the convenience of using Hadoop in the enterprise. Cloudera will achieve its vision of an Enterprise Data Hub by offering what I call 'proprietary convenience' that is intended not to lock in customers but to lock out the competition.

My analysis was that even with hundreds of millions in the bank from Intel that proprietary convenience would not provide a differentiating advantage agains either the data warehouse ecosystem or against Hortonworks and MapR. A billion might not be nearly enough to develop Hadoop into a system that could take large amounts of market share away from Microsoft and Teradata. Because Cloudera is dedicated to keeping as much open source as possible and avoiding lock in, putting that money into making the Apache Hadoop core better only would help MapR and Hortonworks.

When I got on the phone with Olson, I expected that he would throw his Cohiba at me, but instead he had the serenity of someone who was in the 9th inning and 10 runs ahead. Olson was gracious, did not find inaccuracies in my writing, but politely disagreed with my conclusions. Olson's assertions was that Cloudera had the most customers, the most partners, the most money, and would keep doing what it had been doing, now with the help of Intel, and that's that.

As a tech guy who had worked in two companies as a CTO, I was offended that Olson could have so little concern for my analysis of the product. That's when I realized that for Cloudera it is not about the product. It is about the first mover momentum. Cloudera can point to the size of its customer base, its partners, its investment from Intel, and run the old 'Nobody got fired for buying IBM' routine. Olson was happy because I suspect he believes that the massive amount of money from Intel makes this story impregnable.

Lokasi:

Share :