Are object databases “NoSQL” technologies? Part II
I asked the opinion of another ODBMS vendor on the topic of “NoSQL databases”: Luis Ramos who is Principal Systems Engineer at Progress Software.
RVZ: Luis, how do you position yourself with respect to the so called “NoSQL” databases?:
Luis Ramos: We view many of the characteristics of the growing “NoSQL” movement as a market reaction to the realities of present day cloud-based data requirements, where ACID properties are not as important as performance, the bulk of the data’s schema is not as complex, and the corresponding queries are relatively simple. Gone could be the days of complex relational schemas and the DBAs that are needed to maintain and administer them. Similar phenomena have been seen in other areas. For example in programming languages, the reaction against the very complex and error prone C++ led to the popularity of Java.
In many respects, object databases can be classified as “NoSQL” technology. It satisfies many of the pivotal characteristics of today’s “NoSQL” data stores. Object databases have been around since the late 1980s in response to the needs and requirements initially of the CAD market. At that time, the CAD practitioners needed an approach to data management that was fundamentally different than that provided by the relational databases. Consequently, a whole new breed of non relational (object-oriented) databases emerged. Customers from other markets, whose requirements could not be met by SQL databases, followed. Call it the original “NoSQL” movement? We certainly agree with Robert Greene’s stipulation that “one size does not fit all.”
An alternative way to put it is “You can put lipstick on a ‘relational table’ but its still a ‘relational table’”.
The schema-free characteristic that one finds in many “NoSQL” technologies is not entirely new. This is a requirement of many eCommerce applications developed in the 90s. There are object databases that support this nicely, enabling applications to store, manage, and index key/value based data, another key characteristic of “NoSQL” technology. The schema seems very simple but may be challenging to implement in a relational database because the value type is arbitrary.
The horizontal scaling characteristic is another key requirement that object databases more easily supports. Multiple terabytes databases have been successfully deployed. These object database systems have a client-centric (rather than a server-centric) architecture. Data is distributed to the client and queries are performed on the client instead of on monolithic servers. Consequently, the data can be partitioned, replicated, and scaled much more easily without being tied down to the hardware limitations of a single server computer.
So indeed, object database systems could be considered “NoSQL” technologies. They can be utilized either as a persistent store for data as well as a cache.