This is for graphing Cisco IP SLA Probes. For those of you who is not familiar with Cisco IP SLA, it is a probe s/w loaded on Cisco IOS based equipments that you can use to measure round trip times of various protocols/applications on the network. If you have 12.x your router/switch probably have IP SLA capability already present on the equipment, and you just need to create a probe entry. Cisco IP SLA has grown since its introduction in 11.x code (Yup, there are some 11.x code which can do IP SLA, back then called RTR), now it has about 15 or so probe types. Included is HTTP, DNS, ICMP, UDP, Jitter, FTP, DHCP, to name the few. Some you can run with single equipment, some you have to run with two equipments in a client server mode to get more accurate reading on the network health. Check out Emmanuel Tychon's page on more information on Cisco IP SLA:
http://www.etychon.com
I have created a new thread as I do not have access to user ID of original thread. But you can refer to the old thread for historical information on the template:
http://forums.cacti.net/about4136.html
I've added probe specific graph templates since the last time I've updated the templates. Here is the list of graph templates:
Basic - Use this for anything that does not fit in to specific templates (dhcp, tcp connect, etc.)
DNS - shows target DNS query
FTP - shows file size and URL
HTTP - break down of DNS resolve, TCP hand shake, and HTTP, and also shows page size, URL, and HTTP status code
Jitter - Shows source <-> destination positive/negative jitter
Jitter/Error - Shows RTT average value with max min grey band
VoIP Scoring - Shows MOS scoring and VoIP codec
Aside from jitter and http template, the main difference among other templates are the tag lines, which you can take advantage of by importing IP SLA mibs. I will post instructions shortly on how to import the mibs.
5/4/2007
Thanks to Steve W. we have additional jitter template that utilizes standard deviation (SD) instead of simple min/ave/max value. The benefit of graph based on the SD is that outlying data will not pull the average and maximum value out as on the min/ave/max based graph, thus normalizing the graph.
Whilst max jitter is useful to show a worse case, it can be misleading if the SAA sample is based on 1000 packets which is standard for simulating voice i.e. a graph shows a 50ms jitter whereas 999 packets had a jitter of 3ms.
Cisco provide enough data in the MIB to calculate average jitter and standard deviation jitter, so I've created a new graph (in the same style) to do this.
The template package includes quite a few additions (altered data template, altered data query, new graph template, new CDEFs).