10 tips for successful end-2-end monitoring

Traditional monitoring with the monitoring of CPU load, database efficiency or application availability is no longer sufficient for today's complex system architectures. Optimal functioning and high quality of systems and applications can only be achieved with a holistic monitoring system. End-2-End Monitoring which above all also includes the user perspective. However, setting up an end-2-end monitoring process is complex and involves a number of challenges. The selection of adequate technologies, the identification of business-critical processes or the determination of the right thresholds are just some of the tasks.

To avoid mistakes when setting up end-2-end monitoring, IT service provider Consol has compiled ten points:

1. focus on business-critical paths of the application

Running end-2-end tests on a regular basis is resource and maintenance intensive and requires a dedicated test machine, test farm or container setup. Therefore, end-2-end testing should primarily target the core and business-critical paths of an application. 

2. performing end-2-end monitoring on production systems.

Where possible, end-2-end monitoring is performed on production systems rather than on alternative environments with a different setup. This ensures a realistic measurement of the user experience and efficient revenue assurance.

3. determination of the correct threshold values

When an application's performance falls below a certain threshold, its throughput and conversion rate is drastically reduced because the system feels "too slow" for a user's expectations. According to the study "A study on tolerable waiting time: how long are Web users willing to wait?" (2003; Prof. Nah), the acceptable waiting time is just two seconds. Therefore, it is important to determine the right thresholds to maintain business operations. 

4. keep calm

Correctly set thresholds trigger alarms in time for immediate response. In complex systems, load spikes can cause fluctuations in system performance without causing a serious problem. Therefore, an alert may not require an immediate response, but simply an observation of whether the system is stabilizing or whether intervention is required.

5. implementation of global end-2-end monitoring.

If an application is used globally, end-2-end monitoring should be operated in the corresponding local regions and markets. Effects caused by high latency or poor routing efficiency, which strongly influence the user experience, can thus be identified and monitored.

6. isolation from daily business

Data from monitoring checks must not affect day-to-day business. By using identifiable test data and dummy credentials, an impact on business intelligence systems, for example, can be avoided.

7. separation of test and productive data 

Test data and test accounts in production should be clearly identified to separate them from productive data. Transparent, centralized management and documentation of all test data is essential to maintain an overview.

8. encryption of sensitive data

The compliance aspect must not be neglected either, especially with regard to access data for test accounts on production systems. Modern end-2-end monitoring tools have security mechanisms that provide complete encryption of credentials.

9. determination of patterns from the overall results.

A single failed test is not necessarily an indicator of a faulty application. The problems could have been caused by the Internet service provider, the infrastructure, the network or internal services. Instead of checking the results individually, patterns should be determined from the overall results.

10. generate comparable measurement results

Functionalities that are tested depending on conditions can lead to inconsistent or non-comparable results during monitoring if this has a significant influence on the measurement results or if they are only achieved to a limited extent. End-2-end monitoring tests should, if possible, be set up without conditions in order to obtain consistent performance data.

More info is available in the Consol whitepaper "10 Step Guide to E2E Monitoring." to find.