Halfpoint - Fotolia

At the Collaborate conference, optimize BI with Oracle 12c

At the Collaborate 15 conference, Kai Yu wants to share his experience working in BI and as a DBA and how to use Oracle Database 12c to improve BI.

Since the first session he gave at Collaborate 2008, Kai Yu, senior principal engineer and architect at Dell, hasn't let a year go by in which he didn't present at the event. By 2014, Yu reached a grand total of 25 Collaborate conference sessions from over the years. At Collaborate 2015, held April 12-16, Yu will add four sessions to his portfolio.

Among his topics is how to optimize business intelligence (BI) with Oracle 12c In-Memory Database, which first interested him during a project he worked on for Dell to speed up online analytical processing using Oracle BI.

In-memory, he noticed, was the star of Oracle 12c, and, while he could find many papers on how to use in-memory, he did not see "a lot of discussion on how to use 12c with BI," Yu said.

Yu's background is in working with databases, but as the DBA for a BI department at Dell, he picked up BI experience on the job. "I wanted to share that experience," said Yu. "Most people just talk about [Oracle Database 12c] only in the database, not specifically how to use it in BI." Yu sees Database 12c's in-memory feature as an opportunity to speed up the normally slow pace of BI reporting.

As a DBA monitoring BI reports, Yu sees the huge table joins that cause massive slowdowns. Since in-memory is designed to speed up queries, he hopes it can help his colleagues in BI. Columnstore indexes, introduced with Oracle Database 12c, also show promise for speeding up BI reporting. Columnstore is designed to work best with analytical queries like the ones needed by BI professionals. When using a row base, analytical queries have to scan many rows to find the data they need. With a column base, they can find the data they need quicker.

Yu wants to see in-memory speed up reporting, and Dell is working toward using in-memory. However, Yu said, a lot of DBAs aren't on Oracle Database 12c yet. So, his talk at the Collaborate conference will cover not only his experiences, but also Database 12c's BI capabilities, the best features for DBAs and whether in-memory will be good for the users' databases.

Another chance for Yu to share his experiences will come during his session "Design and Implement Your Own Self-Service Enabled Private Cloud with Oracle EM12c." Dell is connecting all of its private storage together and putting infrastructure on top of it. With the cloud pre-configured, if any departments need service, they can request it on the Oracle Enterprise Manager and create the database themselves with no need for a DBA or sysadmin. Everything is pre-built and, Yu said, it takes from 10 to 20 minutes to create a virtual environment and database. Yu plans to talk about the Oracle technology behind this feature and demonstrate how to build a private cloud.

Yu will also tackle storage as a database performance issue in his presentation "Implementing Oracle Database12c's Heat Map and Automatic Data Optimization for Database Storage Performance and Cost." There are two kinds of storage: fast and slow. Naturally, fast storage is more expensive than slow storage. So, to save money, companies are only putting their most active data into fast storage and sending the less active data into slow storage. The trick is how you determine which pieces of data belong in each storage option. Yu's answer is automatic data optimization (ADO) and a heat map to track how often data is used. Frequently used data is stored in the heat map. Then, ADO lets the DBA write a rule that if data isn't used within a specified amount of time, then the data should move to slow storage.

Because many companies buy fast storage from one vendor and slow storage from another, the two aren't able to transfer data back and forth. However, ADO and the heat map aren't on the storage layer, but higher up on the database layer. Controlling slow and fast storage from the database layer overcomes that interaction problem. Yu again speaks from experience with this presentation during the Collaborate conference. The engineering team at Dell provides fast and slow storage both internally and to external customers. Frequently, the IT teams Yu deals with don't know how to use fast and slow storage. This quickly became a common question, which led Yu to conclude that this was an area in which he could raise awareness. Finally, Yu is also the assistant speaker on a fourth Collaborate session, but will take the back seat to the primary presenter, Tariq Farooq.

Next Steps

Learn more about Oracle Database 12c's in-memory function

Find out how to implement Oracle In-Memory Database option for Oracle Database 12c

Dig Deeper on Oracle business intelligence and analytics