In any type of assessment, you need to consider the WYMIWYG aspect (“What you measure is what you get”). Thus, in network and application monitoring, your primary focus should not (only) be on what can be easily measured. Typically it’s things like: Is the server running? Can users access the access points? Do users get a 200 success code in response to HTTP request?
However, what you should ideally be measuring are things such as:
Is the user experience satisfactory? Do we meet business requirements? Can we become more proactive instead of spending efforts on incident response?
In other words you should monitor from an end-user perspective.

Network and application monitoring practices typically suffer from two problems:

  • Too many false positives. With a focus on what we can measure rather than what we need to measure, we generate huge amounts of alarms that need some kind of (costly) follow-up, even if the problem requires no action.
  • Too few accurate problem indicators. In spite of this huge flow of alarms, we fail to verify the most important aspects. Is the user experience acceptable? Do we meet business requirements?

Cloud services and mobility increase the network and application monitoring complexity

In addition, the rapid shift to cloud computing, new sourcing models and new consumption patterns lead to new monitoring needs. Does the Software as a Service (SaaS) provider really deliver up-times as promised? When a server is unavailable, is my Infrastructure as a Service (IaaS) at fault or do I call the network service provider? If end-users complain about homegrown business applications, do they experience flaws in the responsive GUI on their (BYOD) tablets or do we have a back-end problem?

Focus on the user experience

Modern monitoring technologies offer a broad variety of options. We can shift from a focus on protocols and IT components and instead target the user experience and overall business requirements. With agents deployed on clients we can measure what users measure, namely the application performance on their screens. Measuring the key performance indicators that matter from a business perspective we can better control our cloud services, and many other applications and services.

Focus on business criticalities

Legacy monitoring techniques offer great value as long as the objectives are clearly defined. Examples of these include:

  • Network components running with factory settings often represent a severe security hazard. If you simply monitor their availability, rather than their availability in a desired configured state, your monitoring creates a false sense of security rather than an adequate security alarm.
  • Monitoring server up-time is crucial in an IaaS environment. Yet, it may not be the primary concern from a business perspective. A service running clustered across multiple servers is not necessarily severely impacted by one single server going down. The application service itself, however, must be performing without flaws. Thus, application-level monitoring becomes key, whereas alarms related to the performance of individual servers – rather than service clusters – may be counter-productive.

Conclusions

  • A focus on the end-user experience, rather than the health of individual technical components, helps improve performance in IT service delivery.
  • By reducing the amount of false positives in ever-growing streams of alarms, IT organizations save costs and deliver more value to the business.
  • Well-configured monitoring tools help IT organizations identify threats ahead of time rather than spending time on incident management.

At Data Ductus our experienced teams offer operational and advisory services in this field. If you’re interested, don’t hesitate to get in touch.