* Lots of good people leaving - right after they hit the 4 yr mark
* Setting sights to compete with Teradata etc.. means reduced focus on other things that Hadoop has tonnes of potential for.
* Skewed investments favoring certain teams while some teams are left without adequate resources.
* Marketing talks about being a Hadoop company, but focus lately on customer lock-in with proprietary Management Layer and pushing Impala (Which is not really a bad thing maybe, since its a differentiator) and new Tech which does nothing to extend the Hadoop ecosystem anyway.
* Don't really have adequately differentiating / better technology (which cannot be replicated by competitors easily) to warrant NOT going IPO for this long.
* Hortonworks grabbing more developer mind share, and MapR has a substitute for HDFS which justifies its proprietary license by simply being more reliable and faster.
* Engineering Cycles wasted on irrelevant initiatives that don't really push the product forward.