View Printable Version

F09 - Getting the most out of DB2's BLU Acceleration: How to populate columnar tables easily and efficiently!

Session Number: 1296
Track: Big Data and Analytics
Session Type: Podium Presentation
Primary Presenter: Adam Storm [Ibm]
Room(s)/
Time(s):

Laveen => Thu, May 15, 2014 (08:00 AM - 09:00 AM)

Speaker Bio: Adam Storm is a Senior Technical Staff Member in IBM's Toronto Lab. In his 13 years with IBM he has worked in numerous areas of the DB2 LUW kernel including memory management, recovery and storage. Adam is currently the architect for insert, update and delete support in BLU Acceleration.
Audience experience level: Beginner, Intermediate, Advanced
Presentation Category: Data Management, Data Warehousing and Business Intelligence, Implementing New DB2 Releases and Features, Managing Performance
Presentation Platform: DB2 for Linux, UNIX, Windows
Audiences this presentation will apply to: Application Developers, Database Administrators, New Users
Technical areas this presentation will apply to: Database Performance (DB2 for LUW), New Release (DB2 for LUW)
Objective 1: Overview of BLU Acceleration's data model
Objective 2: Populating tables using LOAD
Objective 3: Poplulating tables using INGEST
Objective 4: Poplulating tables using SQL INSERT
Objective 5: Comparing the approaches

Abstract:  Ever since you heard about BLU acceleration, DB2's new columnar parallel vector processing engine for analytics, you've been barely able to sleep. Running queries at lightning speed will allow you to do things you could never have imagined. You can't wait to revolutionize your organization with speed-of-thought analytics. Then it dawns on you, how am I going to populate my BLU tables???

In this talk we discuss the various approaches to populating BLU tables, and their relative merits. We discuss LOAD, INGEST, IMPORT as well as SQL insert and compare their performance, availability and the affect each approach has on compression rates. We conclude by providing a set of best practices and recommendations for when each of these approaches might be most suitable.

For questions or concerns about your event registration, please contact support@idug.org