I am looking for a way to add a bunch of boilerplate text to reports. Right now this is a priority for Availability Reports --- I am having trouble getting managers to remember that the time frame will affect the results.
I have seen the thread about ReportHeader.jpg for logos; however that really would create an ugly situation with these reports.
It seems to me that the engine behind the Report Writer is stone certain to have this feature, so it should be possible to access it.
============
FYI- the current draft text -
The above data reflects the standard operation of Orion Network Performance Monitor (NPM). Availability is not the same as any definition of "outage" or "downtime" that be used elsewhere. Availability is solely the result of NPM trying to ping the target node and recording success or failure as the result. While this is a good general summary of the status of the target node, the details are much more complicated and may present a partially or completely false picture of the operation of the target node. <a href=myserver/avail-details.html>For more details</a> .
--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- avail-details.html
NPM calculates "Availability" by pinging the node every two* minutes. NPM is configured to wait 2500ms (2.5 seconds) for a reply; if no reply arrives in that time the ping is considered to have failed, and the device may be down. NPM will not normally mark a device as "down" immediately if a ping fails; usally there is another two minute period when the device is marked as "warning/unknown" and 'fast' pinging is done ever ten seconds to avoid false positives.
There are two primary items that may distort the data reported here:
1. Network problems between the NPM poller and the node.
"Availability" is normally calculated from the perspective of the individual poller that is responsible for this specific node. If there is a problem on the poller's LAN, or elsewhere in the WAN between the poller and the node, this may affect the results. If there is more than one path between the poller and the node, the failure of one path will usually not appear as down time due to the 'fast ping' behavior mentioned above. If there is no other path, or if al paths have failed, then this node will be reported as down in due course.
e.g.: when the KCWAN Firewalls went down, all nodes outside the KCWAN (Public Access Segment, Front Door Routers, external nodes, etc) appeared to go down from NPM's perspective.
2. Analyzing data for a date range that spans summarization levels.
The results of those pings, the detail data, are kept for seven* days, then it is summarized on an hourly basis. The resulting hourly data is kept for thirty* days, then it is summarized on a daily basis. Any reporting that overlaps between days 0-7, 8-29, and/or 30+ will be wildly skewed. The detail data amounts to 720 records per day; the hourly data to only 24 records per day; and the daily data to a single record.
e.g.: if a node was down all day the day before yesterday, reports would show Availability as:
for Yesterday: 100%
for the day before: 0%
for the last seven days:
what a human being expects: 100+100+0+100+100+100+100/7 = 85.71%
what NPM calculates: (720+720+0+720+720+720+720)/(720+720+720+720+720+720+720) = 85.71%
for the last ten days: (here's the issue)
what a human being expects: 100+100+0+100+100+100+100/+100+100+100/10 = 90%
what NPM calculates: (720+720+0+720+720+720+720 + 24+24+24)/(720+720+720+720+720+720+720 + 24+24+24) = 85.91%
In addition, there are several other items that may impact this reporting:
Changes to or problems with the NPM poller and/or Orion application server and/or SQL database.
Use of the 'unmanaged' option to control reporting of scheduled outages as "down". When used, NPM will skip polling during this period, producing neither failure or success data.
Generating the same report on different dates --- which may change the summarization level of the actual data.
Configuration or operational problems on the target node that affect response to these pings. --------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- References:
