Southern CMG
SCMG Meetings
Interested in presenting at SCMG or sponsoring a meeting?
CMG National

SCMG Meeting Raleigh
October 4, 2013




Fidelity Investments
100 New Millennium Way, Durham, NC 27709

Coming from Raleigh (East):
• Take I - 40 West
• Take Exit 280 on right to Davis Drive
• Keep left at the fork, follow signs for Davis Drive
• Turn Left onto Davis Drive
• Go through two lights and down the slight hill, staying in left lane
• Turn left on Wilkinson Farm Road (may be unmarked)
Coming from Durham: (West)
• Take I - 40 East
• Take Exit 280 on right to Davis Drive
• Keep right at the fork, follow signs for Davis Drive
• Stay right, go through major intersection  and down the slight hill, staying in left lane
• Turn left on Wilkinson Farm Road (may be unmarked)

When on Wilkinson Farm Road, drive until you see the giant parking garage.  Go into the parking garage and find a spot.  The bridge into the Fidelity Facility is located on P2, and a short walk will take you right to the entrance of the building.  There will be a security desk upon entering.  Tell them you are with the CMG Meeting, and be prepared to provide ID.  Someone will be there to escort people up to the meeting room.

This meeting took place on October 4, 2013. Presentation materials will be posted here when available.





Time Session Presenter
Registration, Continental Breakfast and Sponsor Presentation  
Real Customer Experience

Kyle Perrish

Database Performance

Peter Zaitsev 

Discussion on Capacity Planning Tools and Techniques (Mainframe/Midrange/Other)  

Led by Linwood Merritt 

Lunch and Sponsor (Data Kinetics) presentation

Anton Niemand

Some Workload Scheduling Alternatives for High Performance Computing Systems

James McGalliard

NUMA and its Performance Impacts

Claire Cates

ePrivacy Issues and their Potential Effect on Online Data Collection

Anna Long





Database Performance: What Really Matters
Peter Zaitsev, Percona
In many Performance evaluation studies, you will find comparison made in terms of peak throughput or corresponding response time. Yet these metrics are not the ones which well describe how technology would behave in real production systems. In this brief presentation, we will look into why such metrics can be misleading as well as provide framework and principles about performance evaluation which focuses on being able to provide good service in real world production environments. We will use MySQL as example technology, though most of this presentation has broad applicability to data processing technologies.

Some Workload  Scheduling Alternatives for High Performance Computing Systems
James McGalliard, FEDSIM
About 15 years ago, clusters of commodity microprocessors largely overtook custom-designed systems as the high performance computing (HPC) platform of choice.  The design and optimization of workload scheduling systems for these clusters has been an active research area.  This paper surveys some representative examples of workload scheduling methods used in contemporary large-scale applications such as Google, Yahoo, Facebook, and Amazon that employ a MapReduce parallel processing framework.  It examines a specific MapReduce framework, Hadoop, in some detail. 

ePrivacy Issues and their Potential Effect on Online Data Collection
Anna Long, Web Analytica
Websites generally need to collect usage data to be successful, but behavioral advertising and other recent newsworthy events have made many web users question whether they want their activities on the web to be tracked.  This presentation will discuss several projects attempting to standardize how website visitors express their tracking preferences, what data is collected, and which privacy measures are implemented.  These standards could have a substantial impact on websites and how they operate, and they could also provide new capabilities for analytics tools and the analysts who use them.





Lunch: Data Kinetics

How to Use In-Memory Tables to Optimize Data that is Frequently Accessed
Anton Niemand, Data Kinetics/Software on Z 
 In-memory tables provide the shortest path to data. When you have data that is read frequently, and you can reduce the time it takes to access that for each transaction, you achieve significant leverage when you access that data over and over again. This means that you don't have to choose between performance improvement, capacity improvement and cost reduction. For the right data and the right application, the savings are substantial. In-memory tables can be used for one application, can be used to share the same data amongst many applications, or can be used to pass information from one application to another. Because accessing data from in-memory table is so fast, business rules and application specific parameters can be placed in memory, and can be accessed almost as fast as if they were part of the application. This also means that these rules can be changed without modifying application code, providing great flexibility in implementing changes to the business rules. Other ways to use in-memory table to optimize performance of applications are described. A detailed description of how to identify the kind of data that is best for optimization with in-memory tables is provided.