Hello,
my Name is Sascha Schaal. I´m a Developer for a medium sized ISP in Germany and former Nato IS-Admin. I would like make some contributions to Observium, which i think is one of the best approaches to a NMS in the Unix world together with Nagios.
So i have 3 Things for you to offer, which i think could be a usefull addition to Observium.
1. Implementation of WSMan
I could write a Subsystem for Observium to detect Devices and OS´s via the WSMan-Protocol. I think you must have came in contact with a Subset of this Protocol (CIM/SOAP) within the VMware API.
This will require HTTP-POST´s to the Devices WSMan-Listener. These will contain SOAPed chunks of XML which tell the Listener to perform Task´s. You can then iterate through Classes on the WSMan-Provider which contain Perf-Data or Inventory Data and Stati of Devices/or Entitys.
I have developed an easy to use complete OOP Subsystem written in Perl which i would like to convert to PHP for Observium. There are very few dependancy´s like LibXML2 and Data:UUID which i will try to resolve to turn it into PHP, i have some Idea's already.
The best thing is the common Namespaces and Classes in WSMan will make Autodiscovery quite easy with lot of verbose Information to hand over to the User. It goes totaly along with ur approach to a NMS.
Already WSMan capable Devices and OS´s are for example Unix via OpenWSMan, Microsoft Windows and Baseboard Management Controllers like Dell´s iDrac or HP`s iLO.
WSMan has several advantages over SNMP which made the Nagios Dev's plan it for implementing.
my GIT´s on that matter: https://github.com/rage1985/WSMAN--easy https://github.com/rage1985/check_wsman-Suite
Which leads me dirctly to:
2. Discovery, Inventory and Monitoring of Windows OS Metrics
The WSMan ( MS calls it WinRM) is more willing to talk than the SNMP Service on Windows. With the WSMan empowered Observium Nagio´s NS++ Client which has to be installed locally and needs a sh**load of config will look like a joke in comparison. For the Windows Admin WinRM Setup should be piece of cake.
And i know where to find the Interesting Data, and i am willing to share.
3. Statistical Prediction and Forecasting using RRD´s Hold-Winters Algorithm.
I´ve done this several times in my current occupation. Predicting Trends of both Linear and Non-Linear Sets of Data, togehter with Abberant Behavior Detection. I am willing to develope a subsystem for Observium to Predict Trend, Behavior for all the Data gathered in a scalable manner. But therefor i suggest to provide Graphs with fixed timeranges, in a new tab or seperate presantation. I suppose the most users don´t have a clue of the math behind statical analysation of Data. Time is always an important factor in forecasting , u just can´t tell from a month of gathered data what e.g. the traffic next year will look like. So we have to help him out in interpreting his Data. This is done best with sane and easy to read Graphs. I think this kind of Feature will be more than usefull.
I will also help you out understanding the math, the meaning of several values we will stumble upon during developement. And last but not least the Documentation. Here´s is some intro to the techniques and math involved (not from me but good intro): https://www.usenix.org/legacy/events/lisa00/full_papers/brutlag/brutlag_html...
Kind regards,
Sascha Schaal