PennWell Corporation

B2B Media company gains complete visibility into production with AppDynamics

Founded in 1910, PennWell Corporation is a privately held and highly diversified business-to-business media and information company that provides quality content and integrated marketing solutions for industries like oil and gas, electric power generation and delivery, hydropower, and renewable energy. PennWell publishes over 130 print and online magazines and newsletters, conducts 60 conferences and exhibitions on six continents, and has an extensive offering of books, maps, websites, research, and database services.

Challenge: Time wasted troubleshooting

PennWell’s publishing application allows authors to post content to the hundreds of online publications that PennWell supports. To serve PennWell’s customers, this application must perform at a very high level – if a web page is slow or fails to respond, customers file complaints with support. For Ben Hofmann, Manager of Web Development at PennWell, troubleshooting these issues was increasingly difficult due to a lack of visibility into the production environment.

“When we received a complaint from support, the infrastructure team would put the log files and thread dumps in one place for us, and then restart the server,” Hofmann said.

But with gigs of logs to review, it would often take half a day before the architects could begin troubleshooting the problem. Hofmann noted, “The lack of visibility into production and QA meant that we were always in a reactive mode. We were half a step behind our end users because we don’t have the staff to sit around and watch 110 websites all day.”

Furthermore, troubleshooting performance problems often ate up valuable development time. “We’re an agile shop,” Hofmann said. “We usually release code into production once or twice a week. But outages and fixes interrupt our sprints and impact productivity.”

Solution: Visibility into the application layer

Hofmann had already been using Sitescope and Pingdom to monitor website health, but found that these tools didn’t go deep enough into the application to truly provide an application performance management solution. “They only hit our site so often, and only certain pages. They could only tell you if the server was responding, not if there was a problem with the cache or a memory leak, or if there was a problem with a critical business transaction.” For that, he needed visibility into the application layer.

We’ve gained immediate visibility into how our code works in production.

Hofmann first heard about AppDynamics from one of his developers, who had started using the company’s free java performance download, AppDynamics Lite. After a few weeks of using Lite, Hofmann’s team decided they needed the expanded capabilities of the Pro product.

Benefits: Less firefighting, more productivity

With four agile development teams, PennWell’s production environment is altered once or twice a week by new code. Before they acquired AppDynamics, their code drops were constantly interrupted by outages and recommendations for fixes. Today, however, Hofmann’s teams don’t get interrupted nearly as often to put out fires.

“Hofmann said, “AppDynamics gives us access to the log files without a System Administrator pulling them and placing them somewhere for us. We never used to have the visibility into our production environment to know if something was about to bring our application down. With AppDynamics, we’re much more proactive—now we can look at the error rates and slow rates and see a problem building. Then we can call infrastructure and say, ‘It looks like this publisher is in trouble.'"
Recently, one of Hofmann’s developers discovered a problem with PennWell’s RSS feed components. “It was traversing a large amount of data it didn’t need to,” Hofmann said. “The developer quickly drilled down into a slow transaction in AppDynamics and found the error. We were able to catch and fix the problem and no one knew the difference, except us in IT.”

The ability to have visibility into the production environment benefits the entire development team. “The developers really like being able to click through the stack trace,” Hofmann said. “The surface error isn’t very helpful, especially in Java – a null pointer can mean almost anything. Now we can see the line number of code that threw the exception.”

We never used to know if something was about to bring our application down. With AppDynamics, we’re much more proactive.

For Hofmann, AppDynamics allows his team to be proactive in putting out fixes, which improves the productivity of both dev and ops. “With AppDynamics we no longer have to ask for the logs from the production servers. We can see what our code is doing right in our browser, and we can solve problems without customers ever knowing the difference.”