Intial thoughts on cloud databases

Recently there’s been a lot of discussion about alternatives to the standard relational DBMS (note this widely read article).  Some of the feedback comes from OO developers.  There’s a fundamental dissonance between OO and relational approaches, requiring an intermediate object/relational mapping (ORM) layer for OO systems to operate effectively with relational DBMSs.  Others decry the overhead imposed by the relational DBMS, seeking a way to store and use massive datasets for applications needing fast response.  In response a number of vendors are now providing SaaS-like database services in the “cloud”: open source, lightly structured data services provided via the internet, capable of storing and delivering large data stores for high availability, fast response applications.

Vendors have only recently arrived on the scene with these offerings, but according to Kevin Hazzard, CapTech architect, this was inevitable: “In my own writings and talks, I sometimes call databases an unfortunate consequence of history. This sentiment isn’t borne of any sort of animosity toward databases. On the contrary, I can’t imagine doing what I do without the services that databases provide. However, I mustered just enough intellectual honesty to admit that if I were able to give a handful of 4Gb DIMMs to the creators of the ENIAC computer at University of Pittsburgh in 1948 and if I were smart enough to show them how to use that memory effectively, the database as we know it today would simply never have evolved. Instead, databases would have become an integral part of the software development model. The database would be the file system with some automatic hashing and locking capabilities.

“Well, that sounds an awful lot like Amazon’s SimpleDB, Google’s BigTable (and Hypertable), Microsoft SQL Data Services (in the cloud), and other emerging DB platforms.  The ORM trend is also pushing developers toward these models by providing language extensions for query comprehension, query optimization and parallelism in the application domain. Is there a chance that pre-relational or post-relational database engines will eclipse relational databases in the marketplace in the near-term? No way. “

Still, Data Management professionals should be on the lookout for bringing competitive edge to their BI environment with these new approaches – but carefully.  According to CapTech DM/BI specialist Ashwini Kumar, these new database approaches are “a good fit for simple data structures, structured data, and applications with queries using procedural codes. Also, they may be good fit for customized applications. However, Business Intelligence/Enterprise Information Management type applications are more dynamic in nature, especially in exploring or mining data.”  

So where to from here?  Over the coming months we’ll watch as these new database technologies emerge into wider commercial application and lessons-learned emerge.