Where Did NoSQL Come From?
I enjoyed listening to a podcast today of an interview with Jonathan Ellis , who is the project chair for the open systems NoSQL data engine, Cassandra. During the interview, Jonathan related different events causing companies to move away from traditional relational data stores. This is not to say that NoSQL data stores didn’t exist prior to this movement. It does emphasize some of the events that have caused businesses to move away from traditional data storage solutions.
One example shared was Ebay who continued to use Oracle as a data platform until they exceeded the capabilities of the largest Sun computer available at the time. At this point, the need for ACID transactional capability was overshadowed with the requirement to handle transactions at incredible speed. Techniques such as data sharding (storing data on multiple servers rather than a single data store) made the overhead of ACID transactions impossible to maintain.
If you consider the auto-bidding feature of Ebay you can imagine the number of bids that are performed at the close of a sale. Multiply the number of auto bids with the number of auctions occurring on Ebay at any given time and you begin to see the need for higher transaction volumes than previously experienced in many other applications.
Google started out with its own proprietary data store, “Big Table” and later published how they accomplished the monumental task of crawling and indexing the entire world wide web. Map/Reduce integrated with their storage engine resulted in the capacity to process massive volumes of data rapidly, and make that data available at speeds unheard of for a search engine.
Facebook and Twitter have also had their own challenges exceeding the capabilities of a Relational data engine.
Foreign Keys and relationships were no longer implemented in NoSQL data stores, opting for performance and flexibility over optimized schemas or lack of redundancy. Relational constructs did not scale in a highly distributed environment.
But the desire for the ACID nature of a relational database never left. Cassandra is an open systems project started by Rack Space with many contributors attempting to bring together the best of both worlds. While it is not a pure ACID data store, it does have many enhancements to the traditional NoSQL offerings with built in redundancy, or even co-location of data.
Are you integrating NoSQL into your tool kit? I’m finding that more businesses are considering NoSQL not as a replacement to their Relational data stores, but rather as an enhancement. Do you have a hybrid installation? Are you interested in more information regarding NoSQL systems?
Drop me a note at btaylor@sswug.org.
SSWUG TV
With Stephen Wynkoop
Did you check out the two-part interview interview with SQL Azure MVP Ike Ellis?
Watch the Show
Cheers,
Ben
$$SWYNK$$
Featured Article(s)
SQL Server Dump Sniffer (Part 2)
In this post, I shall demonstrate how to use the build output of the Windows Service project to create an installer to install the service on a machine.
Featured White Paper(s)
Structuring the Unstructured: How to Dimensionalize Semi-Structured Business Data
The Business Intelligence industry is paying more and more attention to the ever-growing heap of unstructured and semi-struct… (read more)
Featured Script
admin db – sp_monitor history and historical performance report
Save regular samples from sp_monitor to help you know when it’s time to beef up the hardware on your server…. (read more)