Paxata leverages their enterprise-grade, self-service data preparation application and machine learning to give everyone the power to intelligently profile and transform data into consumable information. Leading global enterprises across financial services, retail, pharmaceutical, healthcare, technology, and public sector industries leverage Paxata to increase analytics speed, quality, and volume, often reducing data migration projects by as much as 75% – 95% while dramatically improving accuracy.
Paxata’s environment runs 70 to 80 hosts at any given time and 25 production servers. Before Instana, they were using Sysdig, a cloud-based APM solution, to monitor their environment. Paxata found that Sysdig was primarily focused on infrastructure metrics such as service impacts, uptime, and disk space. When they needed to analyze any application layer problems, Paxata’s engineers had to jump into their Splunk logs or try to reproduce the problem in their test environments. This made it extremely challenging and slow for Paxata to get to the root of application problems.
“We would routinely spend several days and sometimes more than a week iterating through and trying to figure out problems. We couldn’t reproduce the problem or see how frequently the problem was occurring. Is this a one-time thing? How serious is it? Those things would take us an extremely long time to get to the bottom of.” David Levinger, Vice President of Technical Operations and Support
This challenge was only going to get worse as Paxata was moving to a containerized environment using Docker and Kubernetes. With their existing challenges and upcoming changes, Paxata needed an APM tool that would help them quickly uncover the root cause of issues to reduce their fix time and provide a more consistent user experience. They evaluated APM vendors such as AppDynamics and Stackify, but neither were able to meet the needs of Paxata’s changing environment.
After a thorough evaluation, Paxata chose Instana for its automatic monitoring and root cause analysis capabilities and were able to realize the following benefits.
Easy Install. Immediate observations. Root Cause Identification.
Paxata was able to quickly and seamlessly roll out Instana and immediately observe activity they didn’t know was occuring. For example, certain REST calls returned at arbitrary lengths of time. Using Instana, the DevOps team could see exactly why the calls were slow, which was how the code was interactinag with MongoDB. Instana made it easy to file bugs that included all of the necessary context in Jira. Levenger noted how easy it was to gain new insights using Instana,
“We deployed it, turned it on, and filed two bugs that were super easy for engineering to understand and fix because they knew exactly where the problem was, exactly what the problem was, and exactly how they could fix that problem.”
Based on this quick success, Paxata decided to leverage Instana’s SDK to annotate at the websocket layer.
“The exciting stuff happened. When we integrated the Instana SDK to work with WebSockets we immediately got all sorts of really cool information that led us to file nine serious bugs, three of which were life-changing.”
One of those bugs was related to non-responsiveness in the product. Paxata had seen this a few times in the past but didn’t know why. With Sysdig, they could see strange periods of time where it looked like the application was doing nothing. With Instana AutoTrace, the DevOps team was able to see exactly why and how it was manifesting. The DevOps team set an alert to see how often this was occurring and shared that data with the engineering team for a prompt fix.
“The impact of having Instana is just massive in terms of making it easy for our developers to understand and solve these problems in really well-defined ways, which then leads to them being able to solve more of them, more quickly.”
No Sampling. No Gaps in Observability. Well-Defined Information.
Paxata is a huge proponent of not sampling trace data at scale. Thus, they fully realize the value of Instana’s “no sampling” distributed tracing approach. Other APM vendors Paxata has used and evaluated only sample the data. In Paxata’s view, sampling gives you an illusion that you understand what’s going on. The fact that Instana does not sample data was a huge selling point to Paxata. According to Aaron Witt, Performance Engineer, this is massively important to them as they would not have proactively found a number of bugs had they been using tools that sample data without them first becoming major issues that impacted their customer’s user experience.
In addition to not having gaps due to sampling, Instana also provides source code visibility correlated with the application and infrastructure data to make it easy for Paxata’s developers to identify problems. Levenger added that “Knowing that Joe had a bad experience is one thing, but knowing that this service took 20 seconds interacting with MongoDB, with this query, that was spawned with this code, is another thing. I mean, that’s as actionable as it gets. Like here JIRA here’s this link, give it to a developer and they turn it around astronomically quick.” Instana has enabled Paxata to see the “why” of what went wrong so they could actually fix the problem so it wouldn’t happen anymore. As Levenger and Witt noted, they’ve always been good at restoring service, now they’re able to see and fix the underlying root cause so the issues don’t come back.
With the adoption of Instana, Paxata has complete visibility into their containerized applications. Features such as continuous discovery, high fidelity data, and automatic root cause analysis have given Paxata’s DevOps team a new found confidence in the decisions they make daily to support their customers and their business. They can now see every change, every deployment, every container – discovered, analyzed, and understood in seconds, and automatically identify root cause at any layer of their technology stack to keep their applications running efficiently and smoothly.