Core APM features aren’t going to be enough anymore. Market-leading apps are going above and beyond, setting new standards that others have to match.
Every time any web app, whether it’s aimed at consumers or the enterprise, hits a new performance peak, that sets the baseline for user expectations. It may not be fair or reasonable, but that’s how the market operates.
When someone uses an app for personal reasons and it is a positive experience, they want the same kind of experience for their business apps. That’s one of the drivers of shadow IT right now. People go with the most useful, most productive app that helps them get through the day, even when that app is not authorized by the IT department at work.
Surprisingly, business leaders are embracing shadow IT. Ninety-two percent said that shadow IT apps have made them more productive, and 31 percent said that user-discovered apps have saved at least $10,000 for their organization.
The point is that application performance management (APM) solutions now command more attention than ever before for any company that produces web apps. Your competition is not just others app makers in your vertical. You are being judged against the performance level of every other app in the marketplace.
Many people are familiar with APM’s basic metrics, such as Requests per Minute, Average Response Time, Error Rate, CPU Utilization, Memory Utilization, Disk I/O and Network I/O. Although these are absolutely essential, they are will not be enough to distinguish you in a crowded app ecosphere.
Here are three types of often-overlooked APM features that are helping app makers set new benchmarks for the next generation of apps.
1. Out-of-the-Box or Custom Dashboards
There’s no time to waste on running reports. Anything you need to know on a regular basis should be automated and put on a dashboard for instant reference.
In reality, the metrics you will need are likely to change frequently, so the facility for dashboard creation and management needs to be flexible enough to allow auto-instrumentation without any code changes. The time and resources required to implement frequent code changes can be an insurmountable impediment to optimal APM functioning in most customer environments.
The leading APM solutions now provide a custom dashboarding facility that allows DevOps personal to define ad-hoc charts and graphs of related metrics on the fly. That’s how DevOps teams stay ahead of moving targets in app performance benchmarking.
2. Broader Language Support
There is always a trade-off between depth and width of language support. It’s impossible to widely support every language as fast as they appear, and if a language is not supported deeply enough, it’s better if the APM solution doesn’t claim to support it all. On the other hand, you have to cover all the most important bases because without broad language support, DevOps teams will have huge gaps in their ability to trace down issues. Visibility is key within modern APM, so not having visibility into a distributed request will result in blind-spots in your application monitoring. Having a basic level of visibility, such as latency, into distributed applications will provide at least a starting point for DevOps teams to triage performance bottlenecks.
It’s common for APM solutions to streamline auto-instrumentation for languages like Java, .NET, Python, C/C++ and Ruby. More advanced deployments handle language support for alternatives like Node.js, Scala and Go. However, there are very few APM solutions on the market that provide out-of-the-box auto-instrumentation for popular web servers such as Apache and Nginx. That can give some app makers a crucial advantage. That translates into greater speed in troubleshooting and more comprehensive insights into distributed tracing that others lack.
3. Custom-Defined Baselines/Thresholds
This overlooked feature hold the greatest promise for DevOps teams that are trying to achieve the kind of performance that stands out.
There is a great deal to be gained by user-defined custom metrics, including but not limited to application-specific business metrics, such as conversions, cart abandonment rate and total revenue. While average metrics are useful to gauge the state of most of the cases, percentile metrics (especially the 95th and 99th percentiles) should also be collected to gauge the state of outlier or long-tail cases.
Users are certainly going to use the app in ways that the designers never imagined and under conditions like poor connectivity and intense traffic spikes. Those are the cases where outliers and long-tail can make or break your app’s reputation.
A good example of customized baselines would be triggers and integrations with key alerting tools or third party notification destinations such as PagerDuty or VictorOps. These were created for teams who need to be alerted to warning signs of latency or service interruption but can’t afford to check into every little variation in the performance levels. Both integrate with major APM solutions on one end and issue tracking/customer service platforms on the other.
Your DevOps team should create its own set of custom-defined baselines/thresholds for a malleable routine monitoring temporal events or annotations for critical events, like code pushes or planned outages. It can also help you schedule out subsequent debugging and retrospective sessions. From a user perspective, they can gain a positive impression of your achievement by comparing the performance between releases or detect performance regressions in nightly builds.
The New Normal
Web apps are getting more complex every day as more development teams get comfortable with using microservices and containers. The next generation APM has to be lighter, present a lower overhead and offer better insights into the distributed app performance from the customer’s viewpoint.
The 3 key features of next gen APM solutions that are too often forgotten are:
- Ready-made custom dashboards, right out of the box
- Broad and deep language support for all the most important platforms
- The ability to create your own baselines and thresholds on the fly
While containers have proven incredibly useful in moving apps between environments, they are often so short-lived that they make it harder to report on the condition of the app as whole or recreate the conditions that generated errors in the first place. Conventional approaches to APM won’t be enough to gain adequate visibility into n-tiered apps. You and your team will need to explore the outer limits of your APM’s functionality to push the boundaries of customization and make your app performance the one that sets the industry standard.
If you’re looking for a powerful, yet simple APM solution that provides value at a very affordable price, check out TraceView.
Start a 14-day free trial now!
