Perfbytes

Live from PerfBytes' HQ in Monroe, NC - this is a special 4th of July edition of the show: including round-table discussion, interviews and questions and plenty of interjection from a drunk wookie.

If George Carlin were a DBA he might have said, "A [database] is just a place to keep your stuff while you go out and get more stuff" referring to your application's data. This episode of PerfBytes will review the basics of relational database performance, the core components of a database engine that are critical for good performance and how to avoid a few pitfalls in database design. Oh, you're about to conduct your first database performance test? Listen to a few tips on planning and how to monitor performance in the first few runs. Plus we read some PerfBytian emails in a new segment called the PerfBytes Inbox!

There's this thing between your ears called your brain. We're not joking about this, it's called your brain. If you really want to provide value to your team as a performance engineer, you need to start using your brain. This episode of PerfBytes will help you to understand the basic different types of analysis for performance with a few concrete examples. Mark rants about zombie performance testers, James cracks himself up about Prancercize, and oh yeah, get out that old calculator from college.

Whether it's planned or unexpected, a flood of traffic hammering any system can be a real shocker as the bits and pieces come crashing down around the ops team and the CIO stops by your desk "for a little chat" about what the hell just happened!? We hope this episode will give you some simple advice about how to be prepared technically and organizationally to cope with point load when it happens to you. In other words...this is how to not end up listed on the PerfBytes News of the Damned.

Our good friend and mobile gadget guru John Curtis from Black Falcon Consulting joins PerfBytes for a crazy conversation about mobile performance craziness. We chat about the true impact to your system performance caused by the Bring Your Own Device (BYOD) and what you can do about it. John covers the marketing-simple ABC's for mobile performance and James offers up straight-ahead description of latch conditions due to high-latency mobile traffic. We wrap everything together into a deep-fried taco wrap in a tortilla for this Cinco de Mayo edition of La Radio PerfBytes. ¿Puedes oírme ahora?

It's time for PerfBytes Live! Please join us at STPCON in April 2013 for a live broadcast recording of the PerfBytes podcast featuring your expertly equipped hosts Mark Tomlinson and James Pulley. This show will include our coverage of the STPCON news, current events, industry headlines, extemporaneous speaking from actual performance testers and live Q&A with the members of the audience. You might even 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.

It's a lonely life, as a performance tester in an agile shop. Your work is pushed to the end of the iteration and then they shut you in the lab for four days right before the release. Well, it's time to get performance up-and-out of the lab and build it into the sprint. That's right - you can become a fully-integrated part of the SCRUM as a performance engineer, enhancing each user story with your brilliance about performance. This epsiode of PerfBytes will help you to transcend the lowly life in the lab and engage up-front with developers, product owners and stakeholders to raise their attention to the ever-increasing importance of performance. Where a user story fails to include performance criteria, you'll be there. When a developer fails to run multiple threads of a unit test, you'll be there. Able to construct an integrated performance test into a single nightly test run, it's you to the rescue! We can't promise that they'll give you a promotion, but you might start wearing a home-made super-hero

Production? You remember that system that has to keep running 24x7x365 to keep your customers happy, and employees productive and keep your organization's revenue stream going? Yeah, that one. Did you ever thing that doing destructive load or stress testing on that live system was a good idea? Whether your an external contractor testing over the internet or an internal testing team with no other place to test, sometimes you have no choice but to do your performance testing in production. In this episode of PerfBytes, James and I will give you some tips about how to conduct performance tests in production safely. Sure, we'll try and persuade you not to take the risk - but there are ways to do this, and there are partners that can help you do it right.

There is an old Pete Seeger song that goes like: "Where have all the Performance Gurus gone? Where can they be? (okay, perhaps that's not the exact lyric he wrote, I admit). But when it comes to building a performance team the first thing you've got to master is interviewing prospective performance engineering candidates. What questions to ask? How to structure the interviewing process? And what mistakes to avoid. We interview fellow PerfBytes contributor Scott Moore about his experiences attracting and retaining performance talent. James shares a crazy story about fraudulent resumes and we discuss tips for finding and hiring performance project managers.

Performance Tests, Stress Tests, Soak Tests, Spike Tests, Benchmarks and Baselines...lions and tigers and bears, oh my! In this episode of PerfBytes we'll take a closer look into the definitions for the most common types of performance testing because I know that James loves getting the question: "Hey James, what's the difference between Load Testing and Stress Testing?" You can find out more about why we do the kinds of performance tests we do and the best way to expand the value of your performance testing efforts.

Hey, who turned out the lights? Turn on a LAMP - that's the popular semi-standard configuration of Linux, Apache, MySQL and PHP which is employed across thousands of websites out there in the wilds of internet reality. Even major vendors like IBM and HP are behind the support for this nearly-enterprise-ready stack. Well, it would be totally enterprise ready if you can get it to scale and perform reliably enough to operate mission-critical applications. Join James and I for a sharing of our experiences tuning up LAMPs and illuminating the practices you'll need to succeed. Not that performance tuning wasn't already hard enough to do without proper lighting. Originally posted on Tuesday, February 19th, 2013. Special features include a Perf Rant from Scott Moore (@loadtester) www.northwaysolutions.com

It is not safe to drive a car with your eyes closed. It is also not safe to operate expensive, mission-critical computing systems without monitoring the performance. How much CPU did that query use? How much memory is being allocated/de-allocated? How many bytes per second are being sent and received from the browser? Monitoring the resource utilization of your application is an absolutely essential part of making the necessary calculations to properly test performance. Without those measurements, all you can do is take a guess? Client response time is only one part of the puzzle. Listen to this episode of PerfBytes to learn more about the other pieces of the puzzle and how to put them all together. Original Air Date: Tuesday, February 5th, 2013 8AM EST

Dude, pull your head out of your cloud. You made the leap into the clouds - running new applications, migrating old ones and all the while trying to figure out how to manage and maintain this new shared and uncontrollable infrastructure's performance. In this episode of PerfBytes we'll spend time discussing the in's and out's of THE CLOUD and what steps you need to take to make good on all that elastic scalability and redundancy and cost-saving performance. Mark will share a poem about cloud-based load testing tools. James will have a rant about how you're making the cloud provider's financial dreams come true. Original Air Date: Tuesday, January 22nd, 2013 8AM EST Additional Content: http://www.perfbytes.com/you_cloud

We tuned the databases - for scaling up and scaling out - grids and clouds and no-sql. We tuned the servers - for scaling up and scaling out - farms and clusters and automated deployment. We tuned the network - accelerations and caching and CDNs. What's next? Web Performance Optimization - appealing to developers and designers alike, we are focusing on that last (or maybe first) piece of the puzzle: the client-side. There's been an explosion of tools and techniques around browser optimization and performance analysis, since the first time Steve Souders released the first version of YSlow! Now we have a twitter hash tage #WPO, meetup groups from New York to Seattle, to Frankfurt and Beijing. Vendors are rampantly adding "web performance" capabilities in their existing tools - trying to catch-up with the myriad open-source utilities employed by engineers learning the #WPO craft. This episode of PerfBytes will jump into the party and pass around a wealth of advice and context to understand the Web Perfor

On this day annually it is typical to make a resolution to "never drink that much alcohol ever again" in reference to the previous night's celebrations and the current recovery from a hangover. It's also a great time to make a new year's resolution about system performance to avoid the digital-hangover due to a lack of performance testing. This special episode of PerfBytes is dedicated to A New Year of Performance. We'll hear your stories about 2012 performance - like the time you added 4 indexes to the same entity in the database without updating the stats, or the time you though the new web servers would magically be added to the load balancer VIP - but instead, everything came crashing down around you!? We'll hear stories from load testers, app developers, database engineers, networking engineers and maybe even some DevOps-y gurus about what they're planning to do differently in 2013 to avoid next year's performance hangover. Original Air Date: Tuesday, January 1st, 2013 12PM EST

"Mommy?...where do Performance SLA's come from?" Most project managers and product owners spend the majority of their time studying the functional requirements for the product or application they are building, focusing on user stories that describe HOW an application should work. They forget two other critical definitions for HOW FAST and HOW MANY. This show will discuss the tips for collecting non-functional requirements and performance user stories. How to get started with performance test objectives - how to engage the business owners and stakeholders and how to document and translate those requirements into an effective performance test. Original Air Date: Tuesday, December 18th, 2012 Guests: Perf Rant from Scott Barber (@sbarber) www.perftestplus.com

Look, just because the Matrix was a complete virtual reality (complete with guns and motorcycles and the girl in the red dress), the virtualization fad in our REAL reality isn't quite so sexy. Beneath any gigantic array of virtualized hosts or zones, there are still physical machines and infrastructure supporting your application's performance. Even though I think it's been 5 years since I actually was forced to lay hands physically on a piece of hardware, most of us performance testers are still expected to be just as effective testing on these abstracted virtual systems. In this show the PerfBytes guys will review the challenges, pitfalls and good practices to conduct your load testing with shared resources and consolidated data sources. James gets to do a Perf Rant about running load generators on virtual machines. Original Air Date: Tuesday, December 4th, 2012 Guests: Alon Girmonsky (@alongir) from Blazemeter.com

Imagine yourself as a performance tester. It's less than 1 week away from Cyber Monday - and all your efforts have come right down to the wire. You've recommended more servers and memory for production to handle the load. You've recommended changes in the load balancer configuration to keep the traffic flowing smoothly. You sent your top ten list of SQL queries to the DBA's - hopefully they got the changes committed in time. And now hopefully you have time to call in to PerfBytes and tell us all about it!! We'd love to hear from you folks who have been actively engaged in preparation for what is sure to be the worst day in Internet history. Mark reads a poem about SWAT team performance testing - something to soothe the load tester's soul. James rants about bad configuration for connection pooling. Original Air Date: Tuesday, November 20, 2012 Additional Content: http://www.perfbytes.com/howtocookyourwebsiteforthanksgiving http://www.perfbytes.com/holidayhaiku

Episode #2: the PerfBytes show with James Pulley and Mark Tomlinson - we'll be talking about the leading causes of big website crashes and how to prevent disasters before they happen. As always, please call us with questions about your performance testing experiences and challenges. We'll try our best to help you find the answers, or at the very least we'll be entertaining. Original Air Date: 11/6/2012

This is our first PerfBytes show - call us with questions about performance testing, load testing, stress testing, performance tuning, optimization, performance management, processes and tools. We'll try our best to help you find the answers, or at least we'll be entertaining as we make stuff up. Original Air Date: 10/30/2012