Paper Title and Abstract

Achieving new levels of performance and cost saving, through server and storage consolidation.
Case Study 2016 from a large pharmaceutical company

Stream: DBA

Many organizations are looking to use Virtual technology and consolidate servers running Oracle databases. There are several important reason they are looking to accomplish a consolidation project; savings of the time to manage many servers versus a few, savings of costs around heating/cooling and electrical consumption, savings of licensing of software.

Over the last decade organizations have up graded their servers and increased core/cpu counts to provide more compute power to their Oracle databases. In doing such, they have increased their core licenses counts and thus increased costs dramatically. During the same time the performance of these cores have increased also. This has caused what amounts to a performance glut.

In a review of over 20,000 databases at 1500 organizations we found that 92% of the servers at peak performance were running less than 28% utilization with CPU waits less than 1%. Now, how much utilization is good or bad. In some circles a 100% utilized CPU with no waits at peak would be the best of all worlds. Basically making the full use of your investment in computer without performance wait issues. I however, take a much more conservative approach to this number and allow for more headroom for growth.

In general, trying to consolidate servers running databases, Oracle and others, require a detail analysis of many data points to provide a model of what the real peak performance data points are and where possible bottlenecks can exist. This paper will review the three major legs of a good strategy; Compute, Networking and Storage. I will show how to use AWR reports to review performance and take readings on memory and CPU utilization with various operating systems to create a comprehensive model of a consolidate project. I will show how to use Oracle Virtual Manager to manage and build your consolidated platform and create savings for the organization. 

Pre-requisites:  TBA

Presentation Objectives:  TBA

Speaker Details

Thomas Stutesman has been working with Oracle performance tuning since 1997. He has authored many white papers and taught performance tuning for servers and storage for over 20 years of his 35 year career in Information Technology. He is the architect of several performance tools and has worked with some of the largest customers globally in tuning their environments. His database background included Sybase, Informix, DB2, and Oracle. Presently he is the Global Director of Database Compute Platforms for Hitachi Data Systems, a position he has held nearly 9 years. He manages a database of nearly 2 Million AWR reports from over 4000 consulting engagements he and his team have delivered over the last 9 years. He uses this vast Big Data warehouse to develop performance best practices.