Build resilient systems at scale
28–30 October 2015 • Amsterdam, The Netherlands

performance sessions

13:45–14:25 Friday, 30/10/2015
Steve Workman (Yell)
Slides:   external link
In a world where responsive web design is common, how can the old guard keep pace? Is there still room for adaptive web design in the increasingly responsive web, and can it perform better for your users and your organisation? Steve explores how Yell takes an adaptive approach to keep performance high.
16:10–16:50 Friday, 30/10/2015
Doug Sillars (AT&T)
Slides:   1-PPTX 
In the rush to market, even the best ideas can be implemented poorly. Poorly performing apps lead to unhappy customers, bad reviews, and app uninstalls. In this talk, we’ll look at examples of innovative app ideas that were implemented without thought to performance, or the impact to customers, and we’ll look at ways to prevent such mistakes in the future.
14:40–15:20 Friday, 30/10/2015
Adam Onishi (Financial Times)
Slides:   1-PDF 
Is our fascination with new tools breaking the inherent robustness of the web? Or, can these new tools help progressive enhancement mean something more? My aim is to take a look at the current state of the web, at whether progressive enhancement is still plausible. I also look at how this effects the way we approach the basics of the web, like performance and accessibility.
14:40–15:20 Thursday, 29/10/2015
Simon Hearne (NCC Group)
Slides:   1-PDF 
Web performance is a complex topic, and in our attempts to understand it we’re collecting more and more data. How can we distil this data to make it concise and meaningful in a business context? Simon will explore a number of visualisations beyond waterfall charts, weighing the pros and cons of each, to help you decide what is useful for your business.
16:10–16:50 Thursday, 29/10/2015
Nathan Bower (Zillow)
Slides:   external link
Performance budgets lend a relatively simple safety net that enables designers and developers to build engaging features, while supporting optimal performance or reaching for higher performance goals. I'll discuss how to create good performance budgets, why it's important to have them in place, and tell a true story of how Zillow was able to respond quickly in a case when budgets were exceeded.
11:50–12:30 Thursday, 29/10/2015
Hossein Lotfi (Google)
Measurements are foundational to all aspects of network operations, including diagnosing performance problems, determining reliability, and planning capacity. Metrics allow operators and planners to take informed actions, but how do we ensure that the numbers are meaningful? Are there cases when our interpretation of network metrics is different than what they actually measure and report?
13:45–14:25 Thursday, 29/10/2015
Jakub Pawlowicz (Independent Consultant)
Slides:   1-PDF 
Clean-css is one of the most popular node.js tools these days. We'll take a closer look at how it works and different optimization angles it provides to optimize CSS content. We will also cover how emerging web technologies, HTTP/2 and Service Workers, will affect CSS optimizations techniques in a long run.
17:05–17:45 Thursday, 29/10/2015
Jonathan Klein (Attend)
HTTP/2 is the future of the web, and promises to bring performance improvements, simplified markup, and lower resource utilization on the server. This talk will show you how it works, how to implement it, and how you can get value out of it right now.
13:45–14:25 Thursday, 29/10/2015
Jean-Pierre Vincent (BrainCracking)
Slides:   external link
My daily pleasure as a performance engineer is to use all the tricks in the book to speed up interface rendering and keep it fluid. But performance is actually a UX problem, and sometimes our voodoo is not enough - we have to use real magic tricks. Because perception is key, my favorite UX designer and I review all the non-techie options to speed up our service. Well, to fake it.
17:05–17:45 Thursday, 29/10/2015
Slides:   external link
Finding slow nodes in large clusters is akin to finding a needle in a haystack; hence, manual identification of slow/bad nodes is not practical. The focus of this talk is to present a statistical approach to automatically detect slow/bad nodes, thereby mitigating user impact.
14:00–15:30 Wednesday, 28/10/2015
Philip Tellis (SOASTA), Nic Jansma (Akamai)
Slides:   1-PDF 
Investigating performance problems often requires more than one tool to nail down the problem. In fact, using the wrong tools can often take you completely in the wrong direction. In this talk, we will walk through a real situation we recently dealt with to identify the cause of a performance problem using RUM, Synthetic, and packet captures.
14:40–15:20 Thursday, 29/10/2015
Hooman Beheshti (Fastly)
Slides:   1-PDF 
In this session, we'll examine the challenges around measuring CDN performance and focus on the different methods for measurement. We'll discuss what to measure, important metrics to focus on, and different ways that numbers may mislead you.
11:50–12:30 Friday, 30/10/2015
Philip Tellis (SOASTA), Nic Jansma (Akamai)
Slides:   1-PDF 
It’s hard to tell when a new component of an SPA was requested as a result of an intentional user action or something else. Add on the various ways of building SPAs, ranging from raw XHR to Angular, Backbone, and sometimes all of the above, and you have an interesting problem in traffic analysis. Find out how boomerang measures the performance of SPAs in a way that works for real websites.
14:40–15:20 Friday, 30/10/2015
Slides:   1-PDF 
Performance is a journey, not a destination. Your takeaway from this talk will be a detailed understanding of one company’s impressive performance journey — the inspiration that kickstarted it, the roadblocks it encountered, the tools that made the journey easier, the rewards of sticking it out, and what the road ahead looks like.
11:50–12:30 Thursday, 29/10/2015
Baron Schwartz (VividCortex)
Slides:   1-PDF 
If you're monitoring a lot of data in an event stream (as VividCortex does query traffic monitoring, for example), sampling the stream usefully can be a hard problem. There are all sorts of edge cases and bad consequences. This talk explains how a sketch -- a probabilistic data structure -- turned out to be the answer for us.