Cloudera and HortonWorks: what do I think?

I got asked last week what I thought about the merger of Cloudera and HortonWorks. So, here are my thoughts. Firstly, it looks more like an acquisition. Secondly, while the press release describes the “merger” as being between two companies whose products are complementary, they are clearly much more competitive than complementary. And thirdly, as both companies are loss making, it looks like an opportunity to strip out costs (people) from the combined organisation.

The bottom line is that, in my view, this will be good for HortonWorks shareholders – they get a bunch of money for an over-valued (in stock market terms) company – and bad for everyone else, except maybe the execs (Tom Reilly et al) of the new company. It will be bad for employees, many of whom will lose their jobs; it will be bad for users who will see products and projects dropped or slowed down; it will be bad for IBM and its users who have committed to the partnership with HortonWorks; and it will probably be bad for the future of software development where Cloudera executives, having the whip hand in the new organisation, will favour their own developments over those of HortonWorks, even where the latter are superior. Note that I am not getting at Cloudera executives here: it’s just the way that human nature works.

The big question is whether having one major vendor in this space can be successful – by which I mean be profitable – where two could not? I think this is possible but not guaranteed. The problem is that the story is, and will continue to be, about extra stuff you can add onto the Hadoop platform. But neither Cloudera nor HortonWorks have done this very well. Take SQL on Hadoop: neither Impala nor Stinger are/were nearly as good as some other products, notably IBM’s Big SQL. And the same is true in other areas. For example, consider Spark: who has the traction there? Databricks. Basically, neither Cloudera nor HortonWorks have done a very good job in their chosen market, and there is no particular reason to expect that to change.