The best application performance monitoring solutions (APM) are important for proactively monitoring and managing a software application’s performance and availability. APM’s scope further includes performance measurement for virtually any IT asset that affects end user experience. The sign of the best APM tools are that they detect application performance issues in order to adhere to an agreed-upon service level. In particular, APM is focused on app response times under various load conditions. As part of this, APM also measures the compute resources required to support a given level of load.
According to members of the IT Central Station community, the best APM vendors serve multiple masters. Developers need to understand app performance characteristics in order to ensure an optimal software experience for end users. Business managers and IT department leaders use APM data to help make decisions about infrastructure and architecture.
As applications grow more complex and interdependent, application performance monitoring users express high expectations for potential APM toolsets. Accessibility, manageability and scalability are essential. Users argue that an effective APM tool must give business stakeholders accurate, understandable data while allowing developers to dive deeply into stored data over the long term.
DevOps users want app performance management tools to measure the deep internal transactions that take place inside an application or between integrated system elements. They want APM data in real time, across multiple application tiers, with transparency along the entire application process chain. Some refer to this as “full stack tracing.”
Ideally, APM data should be measured against user experience as a key performance indicator. For example, if a bottleneck is being caused by database latency, users want to understand the root cause so they can fix it immediately. This might require an alerting based on patterns and “baselining.”
Some expect APM tools to enable the discovery of complex distributed application architecture or even microservices and containers. After all, not all application architecture is known at the outset, and it certainly changes over time. Users need APM tools to be proactive whether they are used in dev, test, QA or production environments.
The APM toolset itself should have low impact on application performance. The measurements it takes have to be easy to interpret and place into a business-friendly reporting output. For instance, IT Central Station members suggest that APM tools should offer a predefined customizable reporting capability, with high visibility and a capacity to export and report on large quantities of raw data.
The most important thing to understand about APM products (AppDynamics, Dynatrace, Instana, New Relic) is that they are all about monitoring the code that you wrote and that you are supporting in production that implements one or more business processes that must be reliable and provide a great user experience all of the time.
These products all work by inserting instrumentation into your code that measures how long each function or method in your code is taking, and that captures any errors that occur. In the case of AppDynamics and New Relic, you install their agent into the application run time (this can be automated or scripted). In the case of Dynatrace and Instana, you deploy their agent into the operating system, and the agent automatically injects the appropriate instrumentation into the application run time.
What makes APM different from other monitoring solutions is that APM products directly collect the performance (response time), throughput, and error rate of the transactions that comprise your key business services. There is no API to call to get this information, so the APM vendors all invest heavily in agents that can get this information without creating problems with excess overhead.
APM tools are typically used by enterprises that have mission-critical applications, typically revenue generating, where downtime and/or performance issues result in notable financial loss. For example, an eCommerce website for a large retailer, or a bank’s online banking solution.
In these scenarios APM tools provide deep inspection and transaction tracing across all OSI layers of the application, identifying both physical and logical bottlenecks, as well as application coding issues and/or inefficiencies.
As APM tools can be expensive, they are usually only employed for these critical applications, as opposed to the entire suite of enterprise applications.
A good APM tool provides the user interaction with the application and the application response to the user interaction. Knowing exactly what the user entered and how they navigated to the application feature is a major benefit in diagnosis and mitigation of the issue(s). APM monitoring frees users to do their work while providing real metrics for support.
APM is a very important part of the health of an application. Any application which is already or going to go on production should have tools for a health check-up.
There are different APM tools that we have as of today such as Dynatrace, Splunk, Sitescope, OEM grid and other. All these tools help in monitoring of the applications, servers, databases.
Apart from that, we can do a lot of proactive monitoring so that we are aware of what can break the application and functionality. There are tools such as Sitescope, HP BSM and Splunk where you can configure alerts and on that basis, you can get emails of alerts if something goes wrong.
If there is a slow process in, latency or application and is having errors causing applications to shut down then these tools such as Dynatrace, New Relic, Splunk, OEM grid come handy to do exhaustive RCA.
With my current understanding, through last 10 more years of APM business area, there are still chaos service criteria that APM focused on.
Although a client may not understand clearly, the vendor must consult and guide the right way without manual customization.
We are looking forward to expanding the APM market through a dramatic increase of Microservices and hybrid cloud environment.
APM is like Waze (the navigation application). As long as the car moving it will show you the health of the road but if there are no cars, you can't know anything about the road's condition. So APM is a user experience tool based on transactions analysis, while AIOps Platform provides the SLA based on the infrastructure availability/health.
Nowadays there is a culture for measuring user experience. Keep in mind that an APM solution is not a tool, nor a product, it's a subset of tools measuring (monitoring) infrastructure -> 20% of something (availability, performance) and application -> 80% of something (availability, performance).
In addition to the above, there is a tool for diagnosis purposes. This toy will help you to find out the root cause (bottlenecks) of the event causing bad user experiences (crushes, errors. latency, etc.)
The objective must be clear: speed + quality + security = a good user experience.
Two key points to keep in mind from my experience with APM tools:
1. Avoid tools overload by ensuring you provide skilled people enough time to set up the tool for the specific environment.
2. Don't create agent fatigue. Stop using the latest "magic" tool without considering the failures caused by skipping #1 above.