You're the man.. Thank you for spending so much time explaining this to me. I think I have a good understanding now.
-----Original Message----- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Alex van den Bogaerdt Sent: Monday, November 17, 2008 7:57 PM To: [email protected] Subject: Re: [rrd-users] Absolute values.. ----- Original Message ----- From: "Alex van den Bogaerdt" <[EMAIL PROTECTED]> To: <[email protected]> Sent: Tuesday, November 18, 2008 2:36 AM Subject: Re: [rrd-users] Absolute values.. Sorry, part of it was confusing. Or rather I was confused. I had "--step 1" and steps per RRA entry multiplied by 60 in my mind, but I did not write this down. If you stay with "--step 60", the rate stays 0.5 even when using min or max. When you change to step 1, and have 60 times more steps in each RRA entry, the same amount of time is kept per entry. But when you change average to max (or min) the combined 60 entries will be the maximum (or minimum) rate seen. But, depending on what you are actually monitoring, this may result in fooling yourself and/or others. Why? Because you alter true data into something which is looking like, but not equal to, the truth. Data which should have been zero becomes one, or vice versa. Perhaps you are monitoring an on-off switch or alike. Or a door open/closed, true/false, and so on. It makes sense to have fractions in such a case. The door was open during 10% of the day. The switch was on half of the time. Take a step back and think about what it actually is you want to make visible. Do so keeping in mind that RRDtool will consider numbers to be a rate, "whatever per second". It doesn't know nor care what this "whatever" is, it could even be "degrees celsius times 1 second" which then results in "degrees celsius times 1 second per second -> degrees celsius". It still is a rate. Also think about consolidation. What will happen when you are looking at a graph representing a full year or more? If you frequently see a 1 but also frequently a 0, MAX is going to make your data 1 throughout the year and MIN is going to make your data all zero. If the amount of 1 is rather small, MAX will definately make it visible on the year graph but having one 1 during an hour on a day graph will show up as one 1 during a full day on a year graph. If you do store rates like "the door was open 10% of the day", you (or rather rrdtool) can add such percentages and combine multiple rates into one rate valid during a larger time span. That's what consolidation is all about. You usually don't want to know that the door was opened somewhere in the year. Usually you want to know it was open 10% of the year. And what happens if the data can become more than 1? Can it in your application, once you're out of the development phase? If so, you better start using such rates right away IMHO. _______________________________________________ rrd-users mailing list [email protected] https://lists.oetiker.ch/cgi-bin/listinfo/rrd-users _______________________________________________ rrd-users mailing list [email protected] https://lists.oetiker.ch/cgi-bin/listinfo/rrd-users
