Great news - SPARC T5-8 Delivers Oracle OLAP World Record Performance with near real-time analytic capability

By -

Our SPARC and OLAP development teams have published some fantastic performance figures on their blog: https://blogs.oracle.com/BestPerf/entry/20130326_sparc_t5_8_olap

Using Oracle's SPARC T5-8 server the teams delivered world record query performance with near real-time analytic capability using the Oracle OLAP Perf Version 3 workload running Oracle Database 11g Release 2 on Oracle Solaris 11. The workload uses a set of realistic BI queries that run against an OLAP cube based on a 4 billion row fact table of sales data. The 4 billion rows are partitioned by month spanning 10 years. The combination of the Oracle Database with the Oracle OLAP option running on a SPARC T5-8 server supports live data updates occurring concurrently with minimally impacted user query executions.

This is a great evolution of the development work we have been doing with our in-database OLAP features. The details about "real-time" analytics reminded me of the work with did with BNP Paribas to develop a daily time-series data and analysis solution based around the in-database OLAP features. This customer story was covered in the press and in the January 2007 and April 2008 editions of the OLAP Newsletter:

The BNP Paribas solution stores over 3 billion data values within multiple OLAP cubes across several geographical locations. The solution was based on Oracle 10g utilising Oracle OLAP Option and data access tools for Traders and Quant Analysts. Most of the OLAP calculations, written in DML, have a query time ranging from sub-second to just a few seconds - even for the most complex analytics. The business users access their OLAP cubes through a custom web services front end as well as customised spreadsheet applications. The system supports the retrieval of market data, as well as stored formula, customised functions and complex user generated formula. At the time of the article, BNP Paribas was processing over 10 million rows of data are loaded and continuously aggregating cubes throughout the day.  Overall the system was processing over 20 million data requests per day from over 4000 users.

Cyrus Kapadia, manager of the implementation team at BNP Paribas states, ‘The advantages of Oracle 10g’s scalability, reliability, ease of maintenance, performance and an industry standard analytical query language made it the natural choice to replace our existing system’.

This latest benchmark result from our development team show how we are continuing to develop our OLAP features to deliver groundbreaking performance to support complex data warehouse workloads. In my recent article about in-database analytics I pointed out the importance of having high performance in-database analytics. Therefore, this benchmark is important for OLAP queries but it is also very important in a wider analytical context because it will increase the performance of analytical mashups where OLAP data is mixed and mashed with other key analytical features such as spatial analytics, data mining, time series analytics and/or statistical analytics.  I am hoping to publish a whitepaper on this topic in the next few weeks.

Visit the OLAP Option home page on OTN for more information about Oracle OLAP Option, including how-to videos, presentations, webcasts etc: http://www.oracle.com/technetwork/database/options/olap/index.html

Comments

Post a Comment

Popular posts from this blog

My query just got faster - brief introduction to 12.2 in-memory cursor duration temp tables

By -
Image
Image courtesy of wikipedia This post covers one of the new SQL performance enhancements that we incorporated into Database 12c Release 2 . All of these enhancements are completely automatic, i.e. transparent to the calling app/developer code/query. These features are enabled by default because who doesn’t want their queries running faster with zero code changes? So in this post I am going to focus on the new In-Memory “cursor duration” temporary table  feature. Let’s start by looking at cursor duration temp tables… Above image courtesy of wikimedia.org What is a cursor duration temp table? This is a feature that has been around for quite a long time. Cursor duration temporary tables (CDTs) are used to materialize intermediate results of a query to improve performance or to support complex multi step query execution. The following types of queries commonly use cursor duration temp tables: WITH Clause and parallel recursive WITH Grouping Sets S...
Keep reading

SQL Pattern Matching Deep Dive - Part 1

By -
Image
There has been quite a lot of interest in the new 12c MATCH_RECOGNIZE feature for SQL pattern matching. Therefore, I thought now would be the perfect time to start a series of quick deep dive posts that explain how SQL pattern matching works. Over the coming weeks I will cover the following topics in a series of posts. This is the start of a series of posts based on a presentation that I put together for the recent annual  BIWA conferenc e at Oracle HQ. The Oracle BI, DW and Analytics user community always puts on a great conference and this year was the best yet. You can download any or all of the presentations from this year’s conference by following this link . My pattern matching deep dive presentation started life about a year ago as a post covering some of the new keywords in the explain plan that are linked to pattern matching, see here . It has now expanded to cover a much wider range of topics. Firstly, a huge vote of thanks to my development team (Lei Sheng and Chun-Chie...
Keep reading

SQL Pattern Matching Deep Dive - Part 6, state machines

By -
Image
The obvious way to start this particular post is to pose a couple of simple questions: what is a state machine and why should you care? In general I would say that you don't need to know about or care about state machines. That's the beauty of using SQL for pattern matching. The MATCH_RECOGNIZE clause encapsulates all the deep technical modelling and processing that has to be performed to run pattern matching on a data set. However, there are times when it is useful, probably vital, that you understand what is going on behind the scenes and one of the most obvious situations is when backtracking happens. Therefore, the content covered in this post is a going to be a gently lead-in into my next post where I am going to discuss the concept of “backtracking”  and the dreaded  ORA-30009 error. Let’s start our voyage of discovery…when you attempt to run a SQL statement containing a MATCH_RECOGNIZE clause during the compilation phase we generate a finite state machine ba...
Keep reading