Scalability, Availability, and Performance: Putting It All Together

Scalability, Availability, and Performance -- we want them all. Our users demand them, and large-scale sites can't do without them. While these system characteristics are quite inter-related, there are unmistakable tensions as well; and it is at the point where they all come together that it all gets really interesting. This track will explore proven approaches to designing, building, and operating highly-available, fast, large-scale systems with a series of experts who have done it all.

Day of week:

Wednesday

Location:

Bayview A/B

Host:

Randy Shoup

Presentations

TBD

11:45am - 12:35pm

By:

TBD

Scale and Performance Open Space

5:20pm - 6:10pm

By:

Top 10 - Performance Folklore

10:30am - 11:20am

By: Martin Thompson
High Performance & Low Latency Specialist

Building high-performance systems is tough. Today this is especially tough since a lot of the common wisdom for what makes a system high-performance is misleading at best, and often just plain wrong. This talk aims to expose the myths and folklore commonly found on the web related to building high-performance systems. Martin will cover his updated Top 10 performance myths that even experts building high-performance trading or big data systems fall prey to. Taking an approach of "measure everything" has been an enlightening education into what works, and what doesn't, when trying to achieve high-throughput at low-latencies. This talk highlights the findings of working on systems requiring the extremes of performance and how they truly test any high-performance systems design. Topics covered will include Java, concurrency, operating systems, functional programming, and how hardware really works.

Everything I Learned About Scaling Online Games I Learned at Google and eBay: Scalability at KIXEYE

1:35pm - 2:25pm

By: Randy Shoup
CTO at KIXEYE, former Dir. Google App Engine and eBay Chief Engineer

KIXEYE is a pre-IPO developer of online strategy and combat games based in San Francisco, and has grown 10x in the last 2 years. Funded by Trinity Ventures, we are profitable, and have over 500 employees across the US, Canada, Australia, and the Netherlands. KIXEYE's culture is defined by a passion for high quality, immersive real-time games and a desire to define the next generation of gaming online. We live by the credos Innovation Over Imitation, Quality Over Quantity, and Passion Over Profit.

While the worlds of ecommerce, search, and application platforms might seem as far from the gaming industry as one might imagine, lessons learned in those environments are surprisingly applicable to online games. Real-time games in particular face many of the same challenges faced -- and solved -- by companies like eBay and Google. They are extremely latency-sensitive, are subject to unpredictable growth and scalability curves, and exhibit extremely spiky load profiles. The real-time player experience is critical to the success of the company -- if a game is down or slow, players will leave and never come back. This session will discuss how experiences with large-scale websites like eBay and Google have informed our approach to building, testing, and operating real-time games at KIXEYE.

This session tells several war stories from eBay and Google about performance, consistency, iterative development, and autoscaling. It further puts it all together by connecting those experiences with what we are now doing in our next-generation gaming platform at KIXEYE.

Presentations

Life with a Hybrid Cloud Architecture at Dropbox

4:05pm - 4:55pm

By: Andrew Fong
Dropbox Reliability Lead

Dropbox is used by more than 200 million people to store and access their digital lives from all of their devices. Achieving such scale inevitably comes with a few unexpected challenges, tough decisions, and valuable lessons. In this talk, we’ll detail some of those challenges and the lessons we learned that led us to our current hybrid architecture.

We will also cover some of the operational challenges inherent to using a hybrid data center model, including our thoughts on host management, tool chains, and the unique architectural issues not present in purely cloud or purely physical environments.

Top 10 Performance Gotchas in scaling in-memory Algorithms

11:45am - 12:35pm

By: SriSatish Ambati
Co-founder and CEO, 0xdata

Math Algorithms have primarily been the domain of desktop data science. With the success of scalable algorithms at Google, Amazon, and Netflix, there is an ever growing demand for sophisticated algorithms over big data. In this talk, we get a ringside view in the making of the world's most scalable and fastest machine learning framework, H2O, and the performance lessons learnt scaling it over EC2 for Netflix and over commodity hardware for other power users.

Top 10 Performance Gotchas is about the white hot stories of i/o wars, S3 resets, and muxers, as well as the power of primitive byte arrays, non-blocking structures, and fork/join queues. Of good data distribution & fine-grain decomposition of Algorithms to fine-grain blocks of parallel computation. It's a 10-point story of the rage of a network of machines against the tyranny of Amdahl while keeping the statistical properties of the data and accuracy of the algorithm.

Delivering Performance Under Schedule and Resource Pressure: Lessons Learned at Google and Microsoft

2:50pm - 3:40pm

By: Ivan Filho
Performance Lead at Google

Performance tuning is a specialized skill and one of the most common tasks developers need to face. In this talk we will cover lessons learned during the development and release of several large scale services at Microsoft and Google from the perspective of a performance manager. We will cover success factors and failure points, and how performance should be integrated with product planning, TCO, value proposition, team organization and capability matrix, as well as how the world of online services challenges the traditional algorithms analysis and performance tuning approaches.