2011 in Review: NoSQL Sturm und Drang, While Relational DBs Hibernate

Bookmark and Share

Looking back on 2011, I'm struck by two larger trends in the overall database marketplace. First, most energy and excitement (but not much forward motion) seems to be coming from the NoSQL space.  And second, the major relational database platforms are generating what little energy they can outside of their core RDBMS technologies.

If you kept up on one or more of the better general IT-industry news sources, you probably saw dozens of stories about various NoSQL vendors, spin-offs, and technologies in a single month, compared with perhaps one or two stories in the same period of time covering a traditional RDBMS platform such as Oracle, Microsoft's SQL Server, or MySQL. 

As I first wrote almost a year and a half ago, NoSQL still seems to be driving by its ability to scale with minimal design pre-requisites than for any other reason (see Relational databases can certainly scale up to enormous sizes, but you have to spend a lot of time normalizing (or de-normalizing) the database and configuring the hardware.  NoSQL, in most of its forms, allows you to get started right away and grows by simply throwing more server nodes at the database.

The NoSQL problem for me, and many other data professionals, is that data is an asset that needs to be managed and protected.  This is the primary obstacle.  We're not going to throw our most valuable corporate asset at a fledgling technology.  For some people and enterprises, NoSQL sounds new and cool.  But new and cool translates into "untested and unconventional" for the veteran IT professional.  It's not something to make a bet on. 

Here's one example.  Take any of the more prominent NoSQL platforms - Hadoop or Cassandra or MongoDB - and consider deploying a major new enterprise initiative.  Things that we take for granted on our old-fashioned RDBMS platforms require a lot more angst and analysis.  Where are the industry-vetted best practices? Where are the backup and recovery checklists?  Which hardware vendors have the strongest COEs (Centers of Excellence) and operational research? Who are the top consultancies?  What industry-certified benchmarks exist? Who do I hire?  Heck, it's hard enough to find solid SQL programmers for what's been on the market for more than 20 years! So where will I find quality staff to run this new infrastructure?

And the swirl of buzzwords and hype that accumulated over the course of 2011 made things feel less settled and less reliable.  Thank goodness there are a couple of top contenders in this space; but, the fact that new vendors appear with alarming frequency, again, makes NoSQL seem unsettled.  (Graph databases anyone?  See  The fact that even among the top platforms - say, Hadoop - there are both commercial and open-source versions, along with multiple query languages (Java MapReduce, Hive, PigLatin, and more) just adds to the flux. My head is spinning.

Now, let's contrast the flurry of news coverage about NoSQL with the news in 2011 about the top SQL-based RDBMS platforms, and this sends us into hibernation mode.  Welcome to Yawnville!  Oracle, when it makes headlines, only seems to do so under the context of another acquisition for its applications.  MySQL seems to be mired in politics relating to Oracle's influence.  IBM's DB2 is focused only on carving market share from Oracle's flank.  Microsoft is making a lot of noise, but the news is about its cloud database offerings (SQL Azure) and the strides it is making in the BI (business intelligence) segments of its database stack.  Teradata alone seemed to be attracting some attention for its strides in scalability for the Big Data arena.  If you're looking for innovations in core relational database technologies, you'll find slim pickings over the course of 2011.

So what do these trends mean for 2012?  My hope, given the long gestation period of R&D for new technologies, is that the old school relational database platforms will come out with powerful new releases in the coming year.  There's much talk already about Microsoft's coming "Denali" release of SQL Server, for example.  And, although I never wish harm to anyone, I'm looking forward to a bit more Darwinian action in the NoSQL space so the best can rise to the top and the rest can fall by the wayside.