Moviri at Performance & Capacity 2015 CMG Conference

We had the pleasure of sponsoring and attending the Performance & Capacity 2015 in San Antonio (TX), the 41st US National Computer Measurement Group (CMG) conference.

The Computer Measurement Group is a not-for-profit, a worldwide organization of IT professionals committed to sharing information and best practices focused on ensuring the efficiency and scalability of IT service delivery to the enterprise through measurement, quantitative analysis and forecasting.

What we like the most about CMG conferences is the unique opportunity to get in touch with top capacity and performance professionals, working on different industries and organizations around the globe. It is a pleasure and a privilege to sit together and share experiences and ideas about how to approach actual performance and capacity planning problems! Plus, it is also a great occasion for networking with our customers and prospects, getting to know each other in person and spending some relaxing time outside of the conference hotel.

As Moviri, we have a long tradition of attending CMG conferences to share our best practices and constantly keeping up to date with the advancements of our capacity and performance field.

However, this year was special for us as we not only managed to get a paper approved for this year conference (second year in a row!), but our Stefano Doni had the honor of receiving the conference Best Paper award! We feel especially delighted for this important recognition and wish a huge thanks to all the referees that spent their time reviewing our paper and providing very useful feedback! We’re proud of this international award, we think it speaks for itself about the quality of what Moviri delivers.

Plus, Renato Bonomini managed to deliver an outstanding vendor tool talk on “Business Capacity Modeling with BMC TrueSight Capacity Optimization”, regarded as one of the most appreciated sessions by the attendees – great work Renato!

As usual, we also had a vendor booth which enabled us to present our offerings to US prospects and talk with customers & BMC partners. Thank you, Jon!

Best Paper: Lessons from Capacity Planning a Java Enterprise Application: How to Keep Capacity Predictions on Target and Cut CPU Usage by 5x What is this work all about? Suppose you need to estimate if your user-facing service will be able to cope with the increased traffic of your most important business campaign. The stakes are high, you can’t be wrong or the business will be severely impacted. Chances are that your capacity predictions will fall short and the service will *crash* well before the estimated capacity will be reached. In this talk, we described actionable methodologies that we developed over years of capacity planning business critical Java enterprise applications. Overall, the session has been very well received, we got a lot of questions and interesting feedbacks – thank you! On Dec. 10th 2015, we held a #movinar, a webinar ran by Movirians, on the Best Paper topic. If you didn’t attend the conference and neither the webinar, here is the registration. Enjoy! (If you have any questions, let us know by using the comment box below.)

CMG Session Highlights

As always, CMG is packed with many interesting sessions, selecting and summarizing the best ones is no easy task! Here are two of the most interesting sessions we have attended.

Hadoop super scaling

Outstanding talk by the great Neil Gunther on Hadoop scalability. Hadoop is the most common widely deployed big data framework. Despite its popularity, its performance characteristics are not well understood and the myth of linear (or even super-linear!) scalability is commonly thought among big data developers. Neil shed some light on the real Hadoop scalability and debunked the super-linear scalability myth.

Identify the cause of high latencies in Storage Traces Using Workload Decomposition and Feature Selection

Very good talk by Daniel Myers, a PhD and CS professor who also worked on modeling storage systems @ Google. The proposed methodology allows the identification of the real cause of high latency in storage systems using machine learning. For example, the method can identify which factors (among any) contributes the most to high storage latency, for example high IOPS, high randomness, burstiness and so on. The method was tested against production storage traces collected from Microsoft servers and was able to provide interesting results: nice work!

Conclusions

It’s been a real pleasure to attend this year CMG conference. It was packed with super interesting sessions, we enjoied networking with a lot with customers and partners and were delighted to share an award-winning Moviri-developed capacity planning methodology, hopefully contributing to advancing the capacity planning field with a new actionable approach. Are you puzzled by our findings and would you like to know if the same is happening to your infrastructure and Java applications? Feel free to get in touch with us, we’re here to help you!