There are 2 fundamentally different approaches for monitoring. [1] Poll parameters and strip-chart them displaying some specific duration of history. In this approach data rolls off the end of the strip-chart and is not retained. Only numerical data can be displayed graphically is appropriate for strip-charting. It's strength is that charts are updated at the moment of polling, so all charts are "already made" and it's easy to review many charts at a time (since there's no waiting time at the time you want to view).
[2] Poll parameters and save them in a continuous log file. The strength of this approach is any kind of data can be recorded (numeric or text), and performance can be reviewed for any recorded data (you can look back in time, zoom in on any previous moment). The disadvantage of this approach is that data is "graphed" at the moment of viewing and therefore not as convenient for perusing "multiple graphs at a time" since graphing time must be waited at the time of viewing. Don't confuse long-term strip charting for an equivalent. While you can indeed make a strip-chart display data for a whole "year", typically strip-charts only record the last 400 datapoints over history ... while polling every 5 minutes may provide a history of a couple days retaining all data, charts that display longer term are only retaining average data over longer and longer periods (you lose the actual data polled every 5 minutes). A good tool is one that can do both as appropriate for a given parameter. As far as SNMP, that's just an access protocol. Typically devices provide diagnostics on a web page, or in an SNMP agent. Some devices support both, some support only one. A good tool is one that can poll for data by either access protocol which will serve you best with any collection of equipment. Rich ----- Original Message ----- From: "John Scrivner" <[EMAIL PROTECTED]> To: <wireless@wispa.org> Sent: Thursday, November 03, 2005 11:32 AM Subject: [WISPA] Tracking Signal / Noise / Resends / Etc. I am starting to feel the pains of trying to accurately trace problems in a network that is getting increasingly larger all the time. We have roughly 20 tower locations now. We have a pretty good picture of what we are getting for bandwidth use through our MRTG graphs. We can see when any link completely drops thanks to our What's Up monitoring. We can easily track and isolate peer to peer abuse and such with traffic analysis tools in Star OS and Mikrotik as needed. What we need though is a way to track the other parameters of our network over time. It would be invaluable to me to be able to look over historical records from a hour long window up to a year long window for link integrity information. What I mean is this. I would like to see how a signal level, noise level, number of retransmits, etc. changes over time in a graph format like MRTG does with bits in and out. It would be nice to add other parameters as needed. I know SNMP is supposed to be able to do this but I do not have the slightest bit of experience using that. Is there anyone out there who knows an easy way of getting a rudimentary level of understanding about using SNMP to extract data such as this? Maybe someone has a package they use that provides this information on your network? If any WISPs out there have any answers to this please respond on this list. If you are a vendor and your product can do specifically what I am asking for above then I am sure we would like to hear about that too. Maybe if a few of us buy your products you will consider buying a vendor membership to WISPA also. :-) Thanks guys, Scriv -------------------------------------------------------------------------------- -- WISPA Wireless List: wireless@wispa.org Subscribe/Unsubscribe: http://lists.wispa.org/mailman/listinfo/wireless Archives: http://lists.wispa.org/pipermail/wireless/ -- WISPA Wireless List: wireless@wispa.org Subscribe/Unsubscribe: http://lists.wispa.org/mailman/listinfo/wireless Archives: http://lists.wispa.org/pipermail/wireless/
