Last year Matt Aslett of the 415 Group published an blog post titled “What we talk about when we talk about NewSQL”. In it describes NewSQL as “a loosely-affiliated group of companies … what they have in common is the development of new relational database products and services designed to bring the benefits of the relational model to distributed architectures, or to improve the performance of relational databases to the extent that horizontal scalability is no longer a necessity.”
He then followed-up a week later with a definition of SPRAIN which he defined as:
- Scalability – hardware economics
- Performance – MySQL limitations
- Relaxed consistency – CAP theorem
- Agility – polyglot persistence
- Intricacy – big data, total data
- Necessity – open source
And 451 published a report available for 451 clients, from both the Information Management and Open Source practices (non-clients can apply for trial access). The database landscape diagram is perhaps the most interesting in that it covers a wide variety of databases, placing them neatly into functional and architectural buckets:
What made the blog post even more interesting was the comments made to the article debating the scalability of long time RDBMS Open Source favorite (now owned by Oracle via Sun), MySQL. The Q&A site Quora for example has a question posted reflecting why Quora itself uses MySQL successfully and at high scale. Matt and 451 then followed up this January with an interesting MySQL, NoSQL and NewSQL survey, in which he summarizes the results here. The full slideshare can be Greenplum (acquired by EMC) and AsterData (acquired by Teradata) both started with vanilla PostgreSQL and turned themselves into a shared-nothing, MPP analytical DBMS. AsterData had the additional benefit of adding their flavor of MapReduce in the form of SQL/MR. More recently Hadapt is using PostgreSQL to bridge the gap between SQL and Hadoop, allowing both SQL and MapReduce to be run respectively against the data stored in their separate repositories.
There are also a new breed of databases that focus on Cloud and SaaS, a full list and excellent summary by GigOMs Derek Harris can be found here in the article titled “Cloud Databases 101”.
Finally there are those whose fundamental core is architect from scratch. Let’s face it building a completely new database is hard! And starting from a clean slate brings true innovation. NuoDB is one such company currently in beta. They have a patent for a multi-user, elastic, on-demand, distributed relational database management system, that they tout as everything Oracle is not. VoltDB by Mike Stonebraker (founder of Ingres, father of PostgreSQL and CTO/founder at Vertica before starting VoltDB) is another leading contender in the so called NewSQL camp (incidentally Mike has been quite against the NoSQL movement). RainStor (my current company) by the way is one such database, architected from the ground up to handle Big Data, it is NoSQL in its patented storage mechanisms but it presents a completely relational SQL-92 front end for user access. It can therefore scale like a NewSQL database, but its primary use case is not transactional, but focused on mainly static/read only Big data sets. It also plays nicely with Hadoop running natively on HDFS and supports both MapReduce and PIG access, together with ad-hoc SQL-92 queries.
With the database market seeing more action in the last 2 years than it ever has, and the market estimated at 100 Billion and growing, we will likely see more contenders coming and more contenders forming. Right now there are lots of flavors of databases to choose from solving a variety of use cases, using new and old customized technology. It’s a fun time to be in the database space!