Posted by Tim on July 10th, 2008
I’ve been working on a light weight performance monitoring solution which can be hosted on a LAMP (or WAMP, or OSX =) installation and uses the Google chart API to present data. Data is populated by custom Ruby scripts that either use ssh or perfmon (typeperf) to collect performance metrics from remote Unix and Windows machines.
The intent is to keep the data collection process detailed enough but simple. Presentation of the collected data then uses the API to present charts in a sparkline format.
Sparkline is a type of information graphics, that presents in a simple and condensed way trends and variation, associated with a measurement such as average temperature or stock market activity. These are often used as elements of a small multiple with several lines used together.
The Sparkline is proposed by Edward Tufte for “small, high resolution graphics embedded in a context of words, numbers, images” Tufte describes sparklines as “data-intense, design-simple, word-sized graphics”. Sparklines are intended to be succinct, memorable, and located where they are discussed.
The finished product looks like this:

Other functionality includes the ability to automatically ‘band’ expected data ranges for each metric collected, using a 30 day (customizable) sample that shows the 20 - 80th percentile. This should offer some insight into trends, and leverage off the sparkline format in being able to quickly assess a large range of metrics or performance data.
There’s also a SiteScope plugin, or in fact an emulator, so that if you’re still using LoadRunner you will be able to dip into the same metrics collected by NumbrCrunchr in a similar fashion to SiteScope with minimal configuration changes.
I’ve decided to go open source with this project in the interests of improving and continually updating its capabilities. If you’d like to contribute to the source please go to the project for more information. If you’d just like to see a quick demo, please go to NumbrCrunchr.com.
Posted in Statistics, Analysis, Tuning, SiteScope | No Comments »