From the moment the AWS re:Invent show floor opened Tuesday evening, the New Relic booth was swamped with developers, IT folks, managers, and other cloud enthusiasts looking for more information (and awesome t-shirts).
In addition to ongoing demos, conversations with New Relic experts, and sweet swag, perhaps the biggest draw was a full schedule of short presentations on important cloud monitoring topics. In addition to the slate of New Relic veterans, the booth also featured presentations from a couple of key New Relic partners: Atlassian (makers of HipChat, among other things) and load-testing outfit BlazeMeter.
Atlassian on ChatOps … and better New Relic Integration
Mike Lauricella, director of business development for Atlassian, positioned his company as a champion of HipChatOps, using the company’s HipChat product as a center of real-time sharing and collaboration, and noted that the product’s integration with New Relic has been very popular. But he promised it was about to get a lot better when the company unveils the public beta of its new API on November 3.
Lauricella humorously acknowledged that some current chat implementations could be accused of taking spam out of your email and putting it in your chatbox, but the new API will allow a higher level of automation and integration. Along with other Atlassian partners, he said, New Relic was invited to participate in development and the new update will feature better handling of critical alerts right in the chat flow, along with callouts of “open incidents” on the right sidebar of the screen. Users will be able to “acknowledge” issues and take control of them, he added, while others can follow them to stay up-to-date.
“We’re improving the signal-to-noise ratio,” Lauricella said, giving users key access to critical information right at their fingertips “so you don’t have to ping-pong between apps.” Of course, when issues surface in HipChat (such as Apdex falling below a preset level, as in the example below), you can go directly to New Relic to drill down and take action to resolve them.
BlazeMeter on the 3 steps to better load testing
“Expectations have gotten very, very intense.” That was the warning to developers from Michael Sage, chief evangelist for BlazeMeter. Citing data from Akamai, he said while 49% of users are willing to wait 2 seconds for their page to load, 30% expect action in just one second, and 18% demand “instant” gratification. Satisfy those demands, he said, or risk losing customers to the competition or collecting a bunch of 1-star reviews.
While many companies hold periodic load-testing war games, Sage noted, that approach can’t keep up with modern demands. He offered a 3-step plan for speedier, more agile testing on an ongoing basis.
Step 1: Create the right test scripts. Sage suggested using New Relic to find out where the hot spots are, using real-time results to develop your load-testing scripts.
Step 2: Correlate New Relic and BlazeMeter data. Load tests should address critical KPIs such as average throughput and average response time, Sage said. By using the New Relic API, he added, BlazeMeter can pull in New Relic data such as CPU utilization, and let you jump directly into New Relic to see what’s going on and fix any issues.
Step 3: Test in production (carefully). Testing in production makes some developers and ops folks nervous, Sage acknowledged, but it’s hard to emulate real-world network conditions that can cause latencies and delays. He suggested finding a time with low usage and assemble a team able to track the metrics in real time.
Big thanks to both BlazeMeter and Atlassian (and to Michael Sage and Mike Lauricella) for helping to make New Relic’s AWS re:Invent booth such a success.