Back in the mid 90s I was asked to assemble a framework for managing networks. The framework consisted of several disciplines such as Change, Configuration, Fault, People, Performance and Security Management. At that time it was challenging to find information, Google didn’t exist and therefore you had to rely more on your expertise and network. Fresh out of IBM I had several “Red Books” which were IBMs body of knowledge at that time. I had read several of them before I left and remembered the concepts vividly – note my father worked at IBM all his life, the lingo was embedded in my mind.
Its 1994 and my boss asks me to develop this framework of disciplines based on a whitepaper from IBM. Enjoying a challenge I invested a year into writing whitepapers for the company I worked for surrounding a management framework we could sell to clients. One of the topics Performance Management really stood out when it comes to SharePoint.
Performance Management encompasses the following disciplines:
- Capacity Planning is the process of guestimating network growth and managing the performance impact that occurs.
- Performance Policy Definition is the process of defining performance characteristics and what is required to maintain them. An agreement (Service Level Agreement) may be required outlining the delivered performance.
- Performance Control is the process of monitoring performance against established characteristics.
- System Design is the process of incorporating the needed design specifications required to achieve the desired levels of performance.
- Baselining is the process of establishing and monitoring performance levels for the purposes of delivering a consistent quality of service.
Fast forward to 2011 I remember listening to Spence Harbars presentation in Anaheim with Emer McKenna and while listening I said “This guy gets it”. His credentials aside, most organizations design and build SharePoint environments without a plan for meeting future capacity and performance requirements. They simple design and build an environment for estimated needs, operate it until its begins to fall over then react to Help Desk calls complaining about performance.
Immediately I thought of the Performance Framework I developed back in 94 and applying it to SharePoint as I had done during my time at HP 2000-2010. Specifically, Performance Management involves measuring the performance of key systems. It consists of determining current utilization of applications (e.g. SharePoint) and their components (e.g. SQL Servers, Storage, Network) and analyzing relevant data to discover high utilization trends, setting thresholds and using simulation to determine how the network can be altered to maximize performance.
The goal? Predictive performance, proactive budgeting, improve troubleshooting and determining root cause and its ability to consistently deliver to their Service Level Agreements and minimize risk (Business and technical).
Specific to Baselining, to gauge performance a baseline must be defined. The baseline is used to gauge application (and related devices) performance on normal operating conditions. Therefore the importance of establishing normal operating conditions is critical. The following paragraphs discuss this process in more detail.
Baselining consists of the following steps;
- Define what elements of the application/network is to be baselined.
- Monitor the elements of the application/network defined.
- Set Thresholds, are used to establish a point where action must be taken. This action may relate to a performance or fault item that is critical in nature.
- Monitoring and Notification Tools, are used to alert the console operator as to what thresholds are being surpassed. These tools provide a proactive means of managing the network.
Interested in further reading regarding SharePoint Quality Assuarnce? Read Part 1-3 here , Performance Analysis of Logs toolkit, sample analysis and SharePoint Load Generation Tool. If you’re interested in the Performance Management Document contact me at email@example.com
Interested in another perspective? Have a read