Zenoss Investigates: Why Monitoring Sucks?

September 19 2012 | By | in Monitoring

Let’s face it, a “challenging” career within IT has never been a relaxing walk in the park.  Whether it is the pressure of maintaining legacy systems on a shoestring budget, the challenge of learning new technology, the stress of constant outsourcing pressure, the rewards of carrying a pager (yes, a pager), the pleasure of working a few weekends a month, or being the defacto computer genius for your family and friends, you are the backbone of your companies and your circle of influence is both wide and deep.

If all this wasn’t enough, the very paradigms that have existed for decades are now being turned upside down by cloud computing.  While the term ‘cloud’ may invoke visions of angels, softness and funny faces, the reality is cloud computing is as complex as it is exciting.  Meanwhile, the fundamentals of instrumenting your infrastructure, monitoring, eventing and alerting are more critical than ever.

Here Ye Hear Ye, I have news for everyone; virtualization requires physical hardware!

Unfortunately, hardware manufacturers, independent software vendors and even internal IT organizations have treated IT monitoring as an afterthought.  While virtualization, provisioning and orchestration get all the hype. Monitoring is stuck within a reactive paradigm, built on expensive legacy and proprietary frameworks, and is cobbled together by open source projects, expensive niche tools, or cheap downloadable solutions that offer flash with not much substance.  How are you expected to monitor legacy infrastructure, implement new technology, such as converged infrastructure, and provide a roadmap to the cloud when the very foundation of monitoring is cracking and showing its age?

Therefore, why does monitoring suck?  Is it by design?  Why can we put a golf-cart sized rover on Mars but can’t replace SNMP?  How many elemental management solutions do you need?  Why does every vendor claim to be open but integrations are costly and often fail?  Explain to me why maintenance costs are rising but the frequency of software releases is falling?  Do you really need to learn yet another proprietary scripting language?  Rules? Really, I need to write rules? And my personal favorite, how am I supposed to chose a new monitoring vendor when everyone sounds the same?

To answer these questions, allow me to introduce the 7 Stages of IT Monitoring Grief:

  1. ExcitementFound a new solution
  2. Frustration - Attempted to implement solution
  3. DenialChallenges with solution
  4. AngerThreaten solution provider
  5. DisappointmentThere is no solution
  6. ResolveWe can do this ourselves
  7. HopeSearching for a solution

Where are you within these stages?  Is this an endless cycle?

In the end, monitoring does not suck more or less than any other IT discipline.  The difference is that monitoring must be cross-functional, heterogeneous, and it simply cannot fail.  Monitoring is more than collecting performance statistics, pinging devices for availability, or sending out meaningless alerts to overworked system administrators.   Monitoring is about setting strategic business goals and protecting your brand against failures, outages, and poor customer perceptions.

There are four main tenets to turn monitoring sucks into monitoring luck:

  1. Unified MonitoringA single solution to monitor compute, server, storage, applications, virtualization and cloud computing.
  2. Model BasedThe ability to provide a common language across IT silos by providing a normalized model that fully describes each monitored device.
  3. Event Driven – Alerts are important, but only an event system provides the ability to suppress, de-duplicate, correlate and classify issues.
  4. Extensible Open APIs are important, but the ability to extend the product to provide integrations into 3rd party ISV solutions or new devices is critical.

If you want to hear more about Why Monitoring Sucks or The Seven Stages of IT Monitoring Grief, please join me for an in-depth webinar on October 3rd.