The Hadoop juggernaut keeps on rolling

Hadoop on the march

Back at the end of August I put together a post where I stated that “Hadoop adoption seems to be accelerating“. Now, just a few months later, it is clear that the Hadoop juggernaut just keeps on rolling and that many organizations are adding it to their IT architectures.

Lets take a look at what has been going on (this is in addition to my September post which summarized the latest Hadoop news across Hortonworks, Cloudera and MapR) out there in the busy world of Hadoop.

Hortonworks results

On November 4th Hortonworks released their Q3 earnings. Their Q3 GAAP revenue came in at $33.1 Million up 159% from Q3 2014 and also up from the $30.7 million they reported for Q2. Their Q4 guidance of between $118.6 million and $120.6 million is also pretty healthy as it is above the guidance they provided in Q2 suggesting a strong customer interest they believe they can translate to sales.

On top of revenue they announced 152 new subscription customers which tops the 112 they announced in Q2 and 105 they announced in Q1. At the end of Q2 I counted a total of 556 subscription customers and with the recent announcement they are now up to around 708 subscription customers. As Q4 is normally the largest of any quarter we can probably likely expect them to make it to around 900 subscription customers by the end of the year.

Of course Hortonworks has made some moves. They have entered the Internet of Everything fray with the acquisition of Onyara (more from a previous blog here) and released Hortonworks DataFlow based powered by Apache Nifi. On top of that they have announced a fair bit. Rather than repeat it here please see my twitter stream for links (link filters to just Hortonworks posts) and feel free to follow me for keeping up to date.

There is no doubt Hortonworks continues to demonstrate strong growth for such a young company in a rapidly evolving market.

Cloudera Results

Unlike Hortonworks Cloudera does not release results. This means comparing subscribers is hard. At the end of January 2015 they had 525 customers. I think they are pretty neck and neck with Hortonworks in terms of subscribers. It will be very interesting when they next provide insight into their number of subscribers. Interestingly Cloudera reported revenues of above $100 million already back in early 2015 for their fiscal year so it seems, without being able to verify anything, that they have a better monetization of their customer base happening as with fewer customers than Hortonworks they are claiming almost the same revenue.

There have recently been a series of Strata conferences and roadshows where Cloudera has spoken about things like their One Platform Initiative (aim to make Apache Spark a first class citizen of their platform), RecordService (all about a standard security access model) and Kudu (which seems to be a new way to get the best of both update and query of big data in a simple statement).

I did get to attend one of their roadshows in Munich and the turnout was impressive with many big name companies there. I wrote two blogs focused around that roadshow which I think provide some insight.

They did announce some other things too. As with Hortonworks, Rather than repeat it here, please see my twitter stream for links (link filters to just Cloudera posts).

MapR

Another of the top secret Hadoop vendors :-). Just like Cloudera they do not release actual numbers. With that said they at least shared some percentage numbers. They announced a 160% increase in bookings compared to Q3 2014 (Source) which is almost identical to the year-on-year percentage increase shared by Hortonworks.

While there are no subscriber counts included in their numbers they had around 700 in January so assuming they are adding an average of around 120 per quarter (same as Hortonworks) then they will already be heading towards 1000 customers. I do not believe they are there yet though. My simple reasoning is that I am sure the urge to announce that would have been too great not to do it since it is the next market leadership milestone.

MapR did a number of things over the past quarter including going live together with SAS (aligning with Hortonworks and Cloudera in that regard), bringing in a new President and COO and – my personal favourite – adding free Apache Spark training to their free training curriculum. This and many other things are covered on my Twitter stream (link filters to just MapR posts).

My current view on the Hadoop market

The Hadoop juggernaut shows no sign in slowing down. With around 300-400 new companies buying subscriptions per quarter there is undeniable momentum. Of course the initial subscriptions are often based around small clusters as people experiment. The aim of the Hadoop vendors is not just to keep growing subscriptions but to look to grow investments by existing customers.

To do that organizations need to find that sweet spot of a project which pays for the initial investment and demonstrates the power of Hadoop. With all the tooling that is now available, and the broad array of technologies in the Hadoop eco-system, it is not a matter of if but rather than when companies will go for a big data project of some sort where Hadoop is in the mix either alone or with some other existing big data technology.

As we head towards the end of the year I think we can safely say that by the end of the year we will be close 2500-3000 customers subscribing to one of the major three Hadoop distributions. That is pretty impressive given Hadoop only recently burst onto the scene and despite all the people predicting doom.

Do you have any other thoughts on the market? Am I missing a core piece of data in my analysis? If so share those thoughts in the blog comments.

Mark Torr