Perfbytes

It's time for PerfBytes Live at the STPCON conference in Denver, Colorado! In the unseasonably warm November 2014 this is a recording of the live broadcast of the PerfBytes podcast with your abstractly random hosts Mark Tomlinson, James Pulley and guest host Howard Chorney covering the conference hot topics such as continuous continuousness, performance puzzlers, Denver police swat teams and PRONQ. We have two unusually novel additions to the News of The Damned and as always we give away a pair of PerfBytes shoes to innocent audience members. As supporters of the Software Test Professionals performance testing community, the PerfBytes podcast is powered by creative community-driven contributions and the overwhelming desire to improve performance testing practices across our industry. And pancakes.

After a short summer vacation, your trusted source of performance testing advice, wisdom and humor returns as James and Mark chat about industry standardization and certification, holiday SWAT teams, uncomfortable Kevlar, sacred performance undergarments and microbial shrimp. Welcome Back to PerfBytes!

Hazed and Confused is now the official ice cream for all performance testers and engineers, courtesy of our guest Robert Wescott, who is the author of the recent "Every Computer Performance" book. James and I had a great time speaking with Bob about his experiences in writing the book, translating his expertise into written form and contributing so much to the continued learning that every performance professional desires. James recommends a degree in art history.

Part 3 of the 2014 July 4th PerfBytes BBQ - now we get a little weird...

The second hour of the PerfBytes Live BBQ, we are talking about the introduction of the Performance Manifesto.

It's the 4th of July, so it must be time for PerfBytes Annual meetup and live show broadcasting!

Websites are crashing in the UK, it must be time for News of The Damned. Special guest Seth Eliot joins us for a chat about Zen and The Art of Website Crashing, the spoils of craft brew contaminating the water supply in Portlandia and repeat offender India School Certificate Exam site melts-down again. Vincent Price keeps laughing and websites keep crashing!

Rebuffering, rebuffering, rebuffering...but it's not a surprise that yet another system has crashed in the UK, this time by Sky's not-streaming-enough servers. Cheap airfare, Ricky Gervais, cute dogs, and we welcome Sweden's own Toontrack to their country's first-ever nomination to the NOTD, failing under the load of an anticipated product launch. That guys keeps screaming...still screaming, while we he is punished by the website crashing as Vincent Price keeps laughing. James invites you to have a not-damned week!

Answer me this: how many, how much, averages, standard deviations, frequency distribution or just plain yes or no? It's very hard to know whether your testing effort truly has worth if you don't take a few steps into a few scientific techniques for data analysis. Seth Eliot joins us for a conversation about Data Driven Performance as we learn about the science behind application data, metrics and instrumentation and how to leverage an understanding of the real-world in load testing. We chat about hippo's, the international space station, zero gravity, motorcycle maintenance and mucus creation. This episode is formally sponsored by our friends at www.flood.io.

You journey through a land of both shadow and substance and you can see the curve on the CPU graph is totally going in the wrong direction - a little worse every minute, every user and soon enough the system will just tip completely over. But how can you test for those conditions beyond the capacity of your test lab environment? You gotta estimate future performance based on the results you gather in the lab. Extrapolation might be evil, but not understanding basic scalability concepts is essential to your success in performance testing and engineering. This episode is formally sponsored by our friends at www.flood.io.

PerfBytes frequent listener and long-time performance guru Richard Bishop gives this special report on the recent outages in Great Britain, which is not having such great luck at keeping public-facing websites up-and-running.

It’s time for PerfBytes News of The Damned for May 5th, 2014. Satan once again kicks of a round of impressive stories of website and system crashes from around the world. We pay credit to a drunken ticket sales site, Ted Nugent, Hello Kitty, miraculous tandem web server configurations, and Man United - all of whom have the unfortunate luck, or lack thereof, to be joining us this week! And we learn about typical ASP.NET page design, unprepared lack of load balancing and how a CDN might not solve all your problems. As Vincent Price once said: “Bwaaahhaaahaahaaaaaaa!”

As the spring time flowers blossom and warm sunlight shines upon the meadow, somewhere there's a junior developer struggling to fix a scalability defect in production with not one frickin' clue about how they ended up in that messy predicament. In this riveting episode of PerfBytes we chat about how some organizations end up doing all their performance work in development: as an environment, a phase of work, an organizational role or even just that plain 'ole DevOps thing again. James conjures up Dragnet references, Mark channels Scott Barber and Tim Koopmans lends a Perf Rant about time-to-first-byte vs. WPO! Episode formally sponsored by www.flood.io!

It's time for PerfBytes LIVE! Please join us again at STPCON in April 2014 for a live broadcast recording of the PerfBytes podcast featuring your wildly banal hosts Mark Tomlinson and James Pulley. We are celebrating the one year anniversary of the PerfBytes Live STPCON broadcasting and will include our coverage of the STPCON news, the performance year in review, the News of The Damned, industry headlines, extemporaneous speaking from actual performance testers and live Q&A with the members of the audience. Attendees in-person will be entered into a drawing to win a pair of PerfBytes shoes! As supporters of the Software Test Professionals performance testing community, the PerfBytes podcast is powered by creative community-driven contributions and the overwhelming desire to improve performance testing practices across our industry. And pancakes.

There are even more ticket sales websites crashing in the UK, so it must be time for PerfBytes News of the Damned. In this episode for March 24th, 2014 we talk about Kate Bush’s rockin’ concert dates at the Apollo theater after thirty-five years, a roving underwater shark named Lydia, crowd-sourced searchers for flight Malaysian flight 370, and so many followers of Soul Cycle are disrupted in their progress towards self-actualization. The BBC iPlayer takes a hit and James nearly hurts himself while embracing the inner e-Vil, laughing along with Vincent Price.

Did you know you were hammering the database 1.5 million times per page load? Our friend Petar Puskarich joins us again for this second half of a great discussion about performance diagnostics. We chat about using diagnostics in production specifically with middleware and ops teams, automated deployment of diagnostics configuration, continuous diagnostics and other tips for circumnavigating the politics of having diagnostics across the lifecycle. We cover some tips on working with profiling 3rd party components used by your application and systems internally or externally. And who would imagine that using diagnostics and monitoring together could help you bridge the gap between Dev and Ops. Today’s episode was officially freakin’ awesome as it was sponsored by the colors red and black. Episode formally sponsored by http://www.flood.io, check it out!

Needle. Haystack. Go! Sometimes all we get from our load testing tools is a capture of the response times and throughput, and maybe some minimal resource monitoring if you're lucky. That means to actually see deeply into the application code as it is running, a diagnostics package can open that black-box to you, to find out exactly WHY the system is running slow. As you dig into the system under test you’ll be able to see exactly why the % CPU utilization is so high, and why so many bytes/sec are being pumped through the stack and exactly the SQL statements used by the database. In this Part 1 episode with our guest host and fellow performance guru Petar Puskarich, we'll chat about how to get started with diagnostics, handling politics and objectives, working with third party instrumentation and how Petar's first experiences back in 2001 aren't that unfamiliar to folks today. James is concerned about being abducted by aliens and we have multiple micro-perfrants, all on this deep-dive episode of PerfByte

It’s time for PerfBytes News of the Damned for March 4th, 2014, with our special founding father Carlos Chidiac. Websites are a crashin' like crazy around the world and special applause to the IT folks in the UK who are leading the pack in 2014 with outages! Fanatical figure skating fans, donut-luvin' brits and the on-premise melt-down at Marks & Spencer all join us this week. Bwaaahhaaahaahaaaaaaa!

Automate. Rinse. Repeat. Every day at every hour and every minute there are developers writing code, committing code to the repo, building snapshots, checking out stuff, debugging it, profiling it and generally making things happen continuously. So, why not have performance testing be part of that continuous world? It’s a hot topic in this edition of PerfBytes where we are joined by our fellow pancake-loving co-originator Carlos Chidiac as we debate the value and pragmatics of integrating performance into your continuous integration and continuous delivery process automation. Don’t be scared by James’ Perf Rant about CPT, Agile & DevOps and Carlos who both agree that ‘Speed Kills’ when it comes to your attempts to accelerate performance through automation. Mark shares a few technical tips and we generally advise you to be aware of the pitfalls of trying to promote speed over quality. Episode formally sponsored by www.flood.io, check it out!

At some point after working in the same jobs for twenty years, you start to see a trend in the tools that you use every day. They have similar weaknesses, or similar features for sure, but one thing that you might not spend time considering is the gaps in those tools' capabilities. What if you were calling the shots for the future design and functionality for your favorite performance tool? In this episode we have a good and long chat about those kinds of ideas and features, also discussing some of the top movers and shakers in the industry. Join us again for the birthday episode of PerfBytes, complete with music and party toys! Episode formally sponsored by www.flood.io, check it out!