If you've been asked, "So which one of these solutions should we implement first? And what kind of performance improvement can we expect?" then this course is for you! You will learn how to quickly, quantitatively, and graphically evaluate each of your proposed firefighting solutions. Now your proposed solutions can be objectively evaluated and conveyed to both technical and non-technical audiences. This course contains a rich mixture of lecture, hands-on exercises, and thought-provoking discussions.

What makes this course unique is that every aspect of the systematic performance analysis is quantified and applied to performance mathematics. This provides a number of benefits:

• The entire process can be systematically repeated and taught to others.
• Solutions are more easily identifiable.
• Solution impact can be anticipated.
• The best solutions can be scientifically determined.
• The current and future situation can be conveyed both numerically and graphically.

This industry-unique class was developed by Oracle guru Craig Shallahamer, author of the books Oracle Performance Firefighting and Forecasting Oracle Performance. Craig is a gifted teacher and has a knack for making complex topics come alive while being day-to-day practical. He specifically created this course to inspire DBAs to become true Oracle performance analysts, to encourage DBAs in their quest to thrive in difficult situations, and to empower DBAs with scientific methodologies, diagnosis, analysis, and presentation.

Students are encouraged to bring an AWR or Statspack report to reference during the class. As an added bonus, students are welcome to contact Craig following the course for a one-time free QA of their performance analysis work. All queries, scripts, tools, spreadsheets, and templates referenced are free and downloadable at OraPub's web site.

• How to anticipate the performance impact of proposed solutions on the CPU, IO, memory, and network.
• How performance solutions alter performance mathematics parameters.
• How to perform a complete and scientific database server performance analysis.
• Quickly quantify the performance situation at multiple levels of detail.
• How to integrate Oracle, SQL, the operating system, and the workload into a multi-perspective analysis.
• Be comfortable executing the analysis on systems with an oltp, batch, or mixture type workload.
• Be confident expressing the performance situation verbally, mathematically, and graphically.
• Understand how and why proposed solutions impact users and the overall system.
• How to build solution consensus to motivate change.

• Economic and goodwill loss is minimized because the time from problem conception to solution is minimized.
• Less time is spent discussing possible solutions because they can be scientifically ranked.
• Management can quickly endorse your solutions because they quickly grasp the situation.
• There is much less risk in implementing a performance solution that does not improve the situation.
• Consensus and collaboration is increased because why and how solutions work is crystal clear.

This is the companion and sequel course to our Oracle Performance Firefighting class. We typically offer these two courses back-to-back and offer a multiple course registration discount. If you have attended our firefighting course or read Craig's book, Oracle Performance Firefighting, then you are prepared for this course. Students must be familiar with Oracle's wait interface, architecture, and internals. It also helps to be responsible for Oracle performance and possess a strong desire to eliminate performance solution guessing. This is not an Oracle internals or SQL tuning course, and is not Oracle version specific. A computer with MS Excel is required but usually provided in the classroom. Oracle software is not required.

Database Administrators, Technical Consultants, Performance Specialists/Analysts

Quantifying Oracle System Performance.

This section introduces the core concepts by demonstrating step-by-step how to quantify database performance, how to blend the metrics and performance mathematics, and how to represent the situation both numerically and graphically. Students immediately begin developing a comfort and confidence when communicating using response time terms, numerics, and graphics. Quantitatively representing database server performance allows the natural integration of Oracle performance metrics and performance mathematics resulting in a number of analysis benefits such as problem highlighting, graphical representation, and anticipating proposed performance solutions.

Performing an Advanced Oracle Response Time Analysis.

This section prepares students to take standard performance data related to a session, process, group of sessions, an instance, or the entire Oracle system and then transform the data into a quantitative response time analysis format, which allows for their integration with performance mathematics. More specifically, students learn how to choose the best unit(s) of work, understand their data source options including raw v$ views, Statspack, and AWR, and then how to turn the raw data into a complete and advanced Oracle response time analysis. Also covered are the differences and adjustments needed when working with on-line, batch, and mixed workload focused systems.

Performing an Operating System Situational Analysis.

The operating system capacity and process resource consumption/requirements profoundly impacts any performance analysis. Students learn how to determine and quantify operating system capacity, process resource consumption, and how this feeds into the performance analysis. More specifically, students learn how to creatively use utilization, how to gather Oracle consumption requirements and the associated subsystem (e.g., CPU) capacity, how to gather operating system data from Oracle views, how to determine if there is a CPU, IO, network, or memory subsystem bottleneck, how to integrate the operating system analysis into the broader performance analysis, and finally, how to check that what is observed at the operating system agrees with what is observed from Oracle performance views.

Quantitative Solution Evaluation.

Good performance solutions have a positive impact on our users and ultimately the organization. But how does this relate to technical terms like utilization, throughput, arrival rate, service time, and a response time graph? And how does this affect end-to-end response time or batch run time? This section begins by gently and appropriately introducing performance mathematics and how proposed solutions affect the underlying mathematics. This helps students develop an intuition about why performance changes for better or worse. The focus then changes to why increasing parallelism changes the performance situation and how to mathematically represent this. Then the class works through a number of scenarios about how changes to Oracle, the operating system, or the application workload alter the performance mathematics, the situation, and the response time curve. As an evaluation aid, an MS Excel-based response time curve comparison tool/template is used along with an analysis matrix. Then an enthusiastic discussion ensues about how various performance optimization strategies (e.g., increase latches, tuning Oracle, tuning SQL, adding additional OS capacity, increase the buffer cache, etc.) changes specific response time parameters.

Anticipating Solution Impact Case Studies.

Using real production system examples, our objective is to bring together all the prior course material with the goal of quantifying the performance situation, developing a number of possible solutions, quantitatively analyzing each solution, and then finally to confidently determine the best technical path towards meeting a performance objective. The first case study we do together as a class. For the remaining case studies each student quantifies the performance situation and comes up with a list of possible solutions. Then the class is divided into groups and each group focuses on quantifiably evaluating one of the solutions and presenting their results to the class! It's a wonderful way to bring the entire course together in a very practical and enthusiastic way.

We hope you can join us!

OraPub courses are designed and delivered by Craig Shallahamer, author of: