Let's start with the Cloudera news. Selling the appliance with Cloudera's distribution of Apache Hadoop (CDH3) and Cloudera Manager, Oracle has placed its bet with Cloudera, referencing maturity, reliability, and enterprise quality. Considering the options available to Oracle, this one is considered a low-risk, practical choice.
HortonWorks is gaining momentum, but with less than a year in the market, the maturity of the offering could be a consideration. MapR is proving itself in the industry--however could its relationship with EMC have had an impact on the final decision?
Let's turn to the Big Data Appliance. With a list price at $450k plus support, here is what you get:
- Hardware = 216 CPU cores, 864 GB RAM, 648 TB raw capacity, 40Gb/s InfiniBand
- Integrated Software =Oracle Linux, Oracle Java VM, Cloudera CDH3 and Cloudera Manager, open-source distribution of R, NoSQL Database Community Edition
- Ready to go = pre-installed, pre-configured, tuned, and optimized
This pricing is not too far off from what it would take to build a comparable solution based on a pure Apache Hadoop compute cluster using the scale-out x86 Dell reference architecture. Add the time and skills needed to build, configure, and tune, and you can see how it adds up quickly. Oracle is applying the same engineered systems support model - hardware, OS, software, including Hadoop, is all covered. Because it is a prepackaged rack - the downside is that organizations would buy by the rack - not by the node. Need scale? Add incremental racks - at increments of half a million dollars.
Oracle is accompanying the release with Oracle Big Data Connectors - for existing Oracle Database 11g customers, this provides a convenient bridge between their current investments and a Hadoop-based environment (does not require the Big Data Appliance) for a modest toll.
As we enter into 2012, this announcement speaks loudly to the fact that enterprise IT departments are embracing Big Data and Hadoop and are getting more comfortable with the idea that this will be another foundational component in the data center. When vendors such as Oracle make investments to serve those needs, it is no longer a question of why I need this--but when.