📜 ⬆️ ⬇️

We test not absolutely real (or absolutely unreal) monitoring from New Relic and Appdynamics

Hello, habragraod! Why completely unreal? Yes, because monitoring simulates the actions of real users, but it is performed using a special algorithm from the specified locations. The vendors themselves call this mechanism Synthetics, so further in the article we will try to adhere to precisely this terminology - synthetic monitoring.

New Relic and Appdynamics are direct competitors, are leaders of the Gartner quadrant and implement quite similar functionality. Both vendors have the ability to monitor transactions in applications, they can perform checks on the browser side of the end user, can integrate into mobile applications, and finally have one of the key features of modern monitoring — they can perform algorithmic actions in the web interface of the application.

In previous articles on the basic principles of monitoring business applications and expanding the functionality of Zabbix, we have already touched upon synthetic monitoring and, I must say, if your business grows and the web application serves a large number of clients, Synthetics must have for you. Either he is, or will be. You do not want complaints from customers (and in the worst case - refusal to use your service), right?
')
image

As it is known, the key difference between the two vendors is the presence of the on-premise version of Appdynamics (New Reli works only from the cloud). But in the case of Synthetics, everything turns upside down. Appdynamics can perform verification algorithms only from the cloud: a choice of different continents, of course, except for Antarctica (see the labels on the map). A quick note: this may seem like a crutch, but Appdynamics suggests using three special watchpoints with static IP addresses to monitor internal resources.

image

But with New Relic, you can deploy an observation point both in the cloud and locally at the polar station and check closed corporate resources (for example, to start loan applications in CRM):

image

The next difference between the two solutions is the language used for algorithmization. If in NewRelic, this is JavaScript , which emulates the Google Chrome browser based on Selenium, then in Appdynamics this is Python also based on Selenium, but you can choose from common ones. And here it is worth adding a small nuance: both solutions use Webdriver, but the parameters of this thing (user-agent, screen resolution, etc.) allow you to configure only New Relic, in Appdynamics the default settings are used.

New Relic by default ignores analytics systems , so if you need to include something, use the following algorithm:

Analytic bypass
//Allow Google Analytics scripts to run $browser.addHostnameToWhitelist(hostnameArr: ['google-analytics.com']); //Visit https://docs.newrelic.com $browser.get('https://docs.newrelic.com'); 


In Appdynamics there is no such functionality out of the box, but if necessary, it can be implemented inside your script.

After each transaction step, Appdynamics saves a screenshot of the web page. New Relic saves only the last screen. On the Internet there was one Japanese way, which will be relevant for both solutions. The craftsman from the land of the rising sun describes how to upload files to Google Drive through an API .

In Appdynamics, you can use only standard Python libraries. New Relic also allows you to use the popular Node.js modules .

And the main extracts in tabular form:
FeatureNew relicAppdynamics
Synthetic monitoring from the cloudYesYes
Arbitrary choice of observation pointYesnot
Algorithmization languageJavascriptPython
Webdriver setupYesnot
Boxed ignoring analytics systemsYesnot
Saving screenshots of the screen after each step of the transactionnotYes
Expansion options for algorithms using external libraries / modulesYesnot

Of course, transactional monitoring systems on the market of a wagon and a small truck and determining which one is right for you may not be so easy. In addition to the systems for working with the web, there are also those that are able to run fat clients, conduct OCR-analysis of the interface and issue other pleasures.

Please contact us with questions in the comments. And if the task requires a slightly more thoughtful approach, our consulting, as a bright future, is not far off.

Article author: Anton Kasimov , architect of control systems

Source: https://habr.com/ru/post/313210/


All Articles