APM End-User experience monitoring analysis (bottom)

I. Preface

APM End-User Experience monitoring analysis (above) from the APM end-user product features, usage recommendations, and from the real user experience and simulation performance monitoring, we have a simple sharing.

In this article, we will introduce several new ways to monitor the end-user experience of APM.

Two. Network-based packet capture

The network-based packet capture tool has no effect on the execution of the application and does not require a change of proxy or software, so a large number of application tests are required to derive the desired value. The traffic aggregation layer must be designed in the appropriate location on the network, and a matrix or professional network device must be used to make more granular filtering of the traffic being sent to the monitoring system. The Traffic aggregation network can be used not only for APM use, but also for security monitoring products and network performance monitoring tools, such as OneAPM cloudtest, Newrelic, appdynamic and other vendors are more representative of the network performance monitoring tools provider.

These network-based packet capture products measure response time, detail, error codes generated by the network (such as retransmission, DNS resolution problems, disorderly packet and Nagle delays), A browser-generated error code (such as a request interruption) and an application-generated error code (such as an HTTP error code or other custom error message), some products can also reconstruct and replay what the user actually sees, as shown in the two figure below:

From both the infrastructure and application perspectives, these tools are valuable to the application support team. They can also help business analysts understand the quality of usage and application delivery, such as API monitoring to understand the quality of software delivered by third parties.
4th. Although packet capture technology does increase the security risk of restricting data captured by monitoring software, these technologies are well understood and highly recognized by the mutual network system and other infrastructure management teams, and some traffic aggregation switches allow masking of sensitive data.

As application complexity is enhanced, it is often necessary to use external services from the browser in conjunction with public or personal Web services, combined with content high-speed access and content distribution networks, which leads to a lack of visibility of the network-based approach. Some vendors that provide network-based packet capture can address this challenge by allowing packet capture and analysis on server-side devices, so they are visible when deployed on some cloud-based infrastructure, such as infrastructure as a service (IaaS), software as a service (SaaS), But depending on the traffic situation this can cause overhead and overload of available resources.

At present, domestic and foreign relatively good suppliers such as OneAPM, Newrelic, and so on, they can from the research and development to the launch of the different stages of the corresponding solutions.

Three. Client becomes server

As network-based and other network-centric application delivery methods feature more and more features, the client assumes more processing burdens than before, which requires monitoring of these endpoints. Some methods include an agent-based approach, an embedded agent, or an injected proxy. Some vendors have a solution for these different approaches, but the most common is the injection method, which automatically inserts the code into the Web page, which monitors the time selection and execution of the browser runtime. These technologies add visibility to web-based applications and help us understand the increasingly complex client-side code execution (including, HTML5, and apps).

Manual implementation of these monitoring techniques often complicates the testing of changed software and increases the quality assurance testing cycle. APM vendors are also constantly improving the capabilities of their agents to monitor application servers, inserting such code when processing pages, thereby easing the burden on development and testing teams. These technologies are becoming increasingly important as browsers become a common application delivery platform for the future.

Domestic products so far also Browser Insight this product is more reliable, other manufacturers of similar tools always feel that the dimension is not so rich.

Four. The formation of browser standards

As the current browser uses a lot of manual clocking methods, this prompts the browser vendor to establish a timed application programming interface (API) that is open to software developers. A number of criteria have been developed for this:

Navigation timing--The standard has been widely available in Microsoft Internet Explorer version 9 and above, Mozilla Firefox 7 and later, Google Chrome 6 and above, and Microsoft's Windows phone browser Accept that, as mentioned above, Browser Insight uses this standard.

boomerang--the free-release library uses a variety of timing methods, and it will use the navigation timing specification if supported by the browser.

In addition, Google has quietly released a Google Chrome browser support for a new agreement. The protocol is a high-performance version of HTTP SPDY, which is currently supported by Google services and other internet sites and CDN providers. Hopefully, more of these standards will be used to help us better rely on HTTP, because the core Internet protocol for application delivery is problematic in some cases.

Look for solutions that increase the visibility performance of your browser, which in turn can be used by developers who record performance logs themselves and by APM vendors who provide more detailed performance metrics.

Five. Conclusion

Third-party service providers are not allowed to use agent-based or network-based tools, which makes these technologies seem inadequate. Thankfully, it's possible to improve visibility from a browser perspective based on these application component aggregation. As the end-user experience monitors the market changes, new approaches will emerge that can be used to see the activities and experiences of end users on desktop browsers and on tablets and mobile systems connected via Wi-Fi or data. For network and server tools, it is possible to enhance network packet capture and analysis capabilities through an IT operations analytics approach.

These methods will help us to clear the fog and let the visibility of the clouds be visible.

Cloud Test is a real-time monitoring system based on clouds, which can help you monitor the performance of the website in real time, monitor the availability of third-party service providers such as CDN, DNS, API, transaction monitoring and so on, to realize the timely monitoring and timely alarm. To read more technical articles, please visit the ONEAPM Official technology blog.

This article was transferred from OneAPM official blog

APM End-User experience monitoring analysis (bottom)