Enterprise Monitoring & Reporting - Cleaned

A quick presentation I threw together for a specific organization last year which could easily apply to the needs of many organizations. A few bits have been censored for anonymity.

Transcript

1. Enterprise Monitoring & Reporting Initial Assumptions, Challenges, Strategy and Recommendations Ozzy Johnson - 2013.08.18
2. Strategy 1) Comprehensive inventory of the environment and analysis of any tools and processes currently in use. 2) Creation of a single dashboard to reflect overall system health and identify current, critical issues. 3) Consolidation of redundant tools and replacement of tools which are ineffective or inefficient. 4) Ground-up re-engineering with hierarchical dependencies, trending and scalability in mind. 5) Creation of a feedback loop to drive architectural decisions with empirical system performance data.
3. ❖ No “single pane of glass” to aggregate system health. ❖ Lack of “user experience” monitoring. ❖ Disparate tools applied to monitoring tasks. ❖ Incomplete or missing documentation of system functions and dependencies. ❖ Gathering of metrics without context or meaning. ❖ Costly software licenses, maintenance and support contracts. ❖ Limited scalability of existing tools. Assumptions & Challenges
4. Crafting the ‘Single Pane’ ❖ Effective dashboards cannot be bought, they are hand-crafted by identifying, understanding and aggregating the important data points. ❖ High-quality, flexible, low-cost or free, open-source dashboarding frameworks abound. ❖ * **** **** *** ************* ** ************** ** **** ** *** *********** *** **** ************ ****** ** ******* ****.
5. Grading User Experience ❖ User experience is a composite of every element in the stack from infrastructure to application and content. Full visibility is essential, “silos” and black boxes cannot exist. ❖ Tools are in place to gather this information, but have not been properly interpreted or acted upon. Users should never be the first to notice when a service degrades. Proactive analysis is essential. ❖ * **** ****** ********** ** ***** ***** ** *** *****. * **** ***** *** *** ** ****.
6. Consolidating Disparate Tools ❖ Multiple tools in use by separate departments virtually ensures overlaps, inefficiencies and omissions in coverage as well as licensing. ❖ By inventorying the existing tools for features, strengths, weaknesses and completeness of coverage we will eliminate complexity while reducing costs. ❖ * ** ******** **** **** ** ***** ***** *** **** *** ****** ****** ********* ** ******* *** *********** **** **** *******.
7. Incomplete Documentation ❖ Lack of documentation for the as-built state and users of systems and absence of baseline performance stats ensure that every exercise in troubleshooting must reinvent the wheel. This issue is bigger than monitoring, ** *** ********* ** ******* ***** ******** **** ** *** *** **********. ❖ Documentation is essential, it must be created, derived from existing systems if necessary. ❖ * **** *** ************ ********** *** *********** ** ****** **** *******.
8. Metrics Without Meaning ❖ Existing tools gather large amounts of information in need of context and interpretation. **.*** ****** ******** **** *********** ** **** (******) ****** **** (****) ❖ Addressing this begins with the survey of existing systems to determine relationships, dependencies and set baseline expectations. ❖ Interpreted data will drive informed changes.
9. Working At Scale ❖ Scaling traditional enterprise monitoring and management tools is made difficult by their closed nature, high costs and complicated licensing schemes. ❖ This is a solved problem. Free, open-source software monitors and maintains the largest, most visible systems in the world. ❖ Open-source, scalability and automation *** ******** ** **** *** * **** ****** **** **-**-**** ** *** ***** ** *** ***.
10. Next Steps ❖ Identify the most pressing pain points for immediate integration into an initial dashboard. (*******: ****/**/**). ❖ ****** ****** ** ******** ********** *** ********** ******* ** **** ** **** **************. (*********: ****/**/**) ❖ ******* ************* ** ******** ****** ************* *** ************ (*******: ****/**/**).