NoSQL Movement Gains

There’s growing dissatisfaction with SQL on a number of fronts, but it seems hard to tell if this represents an actual movement or just a loose coalition of groups with slightly aligned interests.

The most prominent members of the “NoSQL” movement are proponents of new data management schemas such as Hadoop and MapReduce, the latter being a new approach to data management pioneered by Google. Vendors that have embraced these approaches include Aster Data and ParAccel.

A less prominent faction of the “NoSQL” movement comes in the form of companies such as Mark Logic, which makes an XML database that supports the Xquery language for accessing unstructured content in preference to a derivative of SQL.

Finally, there is a group from the open source community that is promoting a Couch database that allows users to query documents stored in the database using JavaScript.

While these factions represent the major technology elements of the  backing the “NoSQL” initiative, there are other IT folks lending their support because of their frustrations with database pricing, while others such as Terracotta simply want to reduce the database to a commodity by relying more on memcachedb approaches.

By leveraging the popularity of Google and other famous Web 2.0 companies that have eschewed the SQL database, the “NoSQL” effort is now fashionable than ever.

Every faction in the “NoSQL” movement has some legitimate issues, but as Mark Logic CEO Dave Kellogg points out the challenge with all movements is that they can become reactionary. There will probably always be a need for a SQL database. But that said, a SQL database does not need to be the center of the universe for all data. There are instances where the sheer volume of data or the very structure of the data makes another option more viable.

The real challenge will be getting existing database administrators to realize that.

Comments

Consider the revolutionary new illuminate Correlation Database; this database management system (DBMS) is data model independent and designed to efficiently handle unplanned, ad hoc queries in an analytical system environment. A correlation database uses a value-based storage (VBS) architecture in which all data is indexed -- every value in every field -- and unique data values are stored once and only once, no matter how often the value appears and regardless of its original location. Advantages: no need for database pre-design or upfront requirements definition; a data-driven logical scheme is automatically built as data is indexed during the loading process. In fact, because all information is indexed, the scheme includes all possible data interrelationships (or correlations). New data sources can be added on the fly. A correlation database is also ideal for integration of data from disparate (legacy) databases. No physical schema is needed. Fast time-to-answer. Ultimate ad hoc query flexibility. We invite you to provide us your data challenge - http://www.datainnovation...
There are a lot of different approaches than just the one listed, and many more useful implementations. There are the key value stores, such as Memcache (with support from Gear6), and Tokyo, and Reddis, and Tin. FluidDB is an interesting key value store approach, in that it is completely "public", anyone can add data to the store, and anyone can query it. There are the BigTable systems, such as Cassandra and Voldemort There are the document stores, such as CouchDB and MongoDB. There are hybrid attribute stores, such as AWS SimpleDB.
As further evidence, check out the upcoming sold out NoSQL Live event in Boston this week (March 11). There will be a free webcast for those not able to get a ticket - http://nosqlboston.eventb...

Post new comment

The content of this field is kept private and will not be shown publicly.
  • Web page addresses and e-mail addresses turn into links automatically.
  • Lines and paragraphs break automatically.
  • Allowed HTML tags: <b> <i>

More information about formatting options