Oracle Defends Relational DBs Against NoSQL Competitors
Oracle: NoSQL Isn't Anything New Moving foward to the last 10 years or so, Mendelsohn said, "the latest generation of these NoSQL products came along. They were designed just like the original NoSQL products to solve very simple problems—you can't do things like reporting and analytics with these (NoSQL) databases. They're very developer-intensive, very low productivity. "Amazon created DynamoDB, which was actually a descendant of an Oracle product called BerkeleyDB (from the Sleepycat acquisition in February 2006). Amazon and Google were big customers [of Oracle's]. It's an open source product; Amazon was buying a lot of support for it, and at some point they just took it because it was open source and created Dynamo around that same technology," Mendelsohn said. "I don't know if Berkeley is still in the core of DynamoDB, but that's where Dynamo came from. It was designed for these simple data management problems, like if you were going to tell Amazon, 'I am Andrew Mendelsohn, show me some products that I can buy, or show me a price list.' These were not transaction processing; these were sort of data-publishing applications that were very simple key-value use cases."As far as all the heavy lifting at Amazon, Mendelsohn said, "they still to this day use Oracle relational databases to do all the transaction processing on the Amazon.com Website; they use Oracle DBs for all the data warehousing on the e-commerce site. All the big data, all the analytics is still done by Amazon on Oracle." Certainly, Amazon could use DynamoDB for the big data workloads, Mendelsohn said, "but they don't because SQL is a much more productive way of getting analytics out of a database that it makes no sense to use NoSQL databases. "The end of the story is pretty simple: NoSQL products are very good for very simple applications, where you're pointing data back to the key, you're getting a value back, and it can be a JSON document or whatever it is that people want. Relational SQL databases are really good when you're trying to do more complex workloads, like getting a report back for your business and doing transaction processing," he said. The bottom line, Mendelsohn said, is that NoSQL, which has been around for 40 years, is still a small niche of the market. "Relational databases are pretty much all of the market," Mendelsohn said. "Relational database have the attributes to do everything; NoSQL is limited. You can believe me or not believe me, but if you look at the Gartner market share numbers, it's pretty clear which market is big and which is small."
OracleDB Does the Heavy Lifting at Amazon