Despite a host of new big data innovations, including Hadoop, MapReduce and NoSQL databases such as MongoDB, some...
familiar and trusted technologies hang on -- one being SQL.
That is because the job of uncovering important business patterns doesn't become easier as companies collect data in more varied formats, especially if those new formats don't have familiar SQL support.
SumAll provides services that allow its users to tap into social and Web application performance data. According to Korey Lee, SumAll's chief information officer, the company is looking to use SQL to generate reporting dashboards that help customers identify social media trends and measure their social media efforts versus their industry peers. Twitter, Instagram, Facebook -- these are just some of the many Web applications that are supported.
BitYota's software services "build a mapping layer on top of MongoDB that allows us to query against our data and aggregate it," Lee said.
He said SumAll tracks 100s of billions of social impressions over billions of rows. He characterized MongoDB as very scalable, but added that it can be difficult to analyze and aggregate MongoDB data.
"MongoDB has its native query functionality of MapReduce," Lee said, referring to the programming environment often used to process large data sets with Hadoop clusters and NoSQL databases. "But we found that it doesn't do a really good job with aggregations for us, especially across collections [of data] at scale. BitYota solves that problem."
By such means, he said, SumAll can create analytical dashboards that let customers estimate the return on investment and compare results for marketing campaigns across social media platforms.
"We can help our customers figure out if Instagram is more effective than Facebook for them -- and, even more importantly, track if this is a broader trend across tens of thousands of business users," he said.
JSON and the 'NoSQLnauts'
The need to analyze NoSQL data from sources like MongoDB or Hadoop appears to be growing. BitYota CEO Dev Patel said his company, based in Mountain View, Calif., formed around the need to integrate varied types of data from around the Web, and that cloud data warehouses can help enable such applications.
NoSQL databases, supporting data schema creation after the fact, keep users' options open for incorporating all kinds of data as time goes by, he said. Still, SumAll's Lee and others continue to look to SQL for queries on diverse Web-borne NoSQL data, Patel noted.
"We see a lot of interest in NoSQL related to flexibility, especially in the mobile and Web application space," he said. "But there are many areas where SQL is going to continue."
Meanwhile, the MongoDB database arose, often used as a store for JSON documents. The trend has not been overlooked by established SQL tools vendors such as IBM, Microsoft and Oracle, which have begun to offer various ways of handling JSON. They have customers that, like SumAll, want to apply SQL methods to NoSQL application data.
SQL is a big reason why SumAll went with the BitYota data warehouse service, Lee said: "It lets us use something familiar."
The actual implementation of SumAll's MongoDB version comes via ObjectRocket, a Rackspace U.S. Inc. cloud hosting company that provides MongoDB as a Service. Speed was an important factor in SumAll's software selections, according to Lee, who said the tools employed allow SumAll to query its full store of data on demand at any time.
Watch a video Q&A with MongoDB's CEO
Go behind the hype on NoSQL databases