Calculating this would require some major changes, and I'm not convinced it's actually something useful to calculate.
 
As a workaround you can just put each port as a separate bill and add them together.
 
What your provider is doing is very dishonest. The whole purpose of the multiple-port billing system is for circumstances such as these, where you buy a multiple redundant ports for the same service. You should obviously be billed on the 95th of the aggregated traffic, not on multiple separate 95th calculations added together.
 
If I were you, I'd change provider.
 
adam.
 
------ Original Message ------
From: colin.barker@bluesky.co.uk
To: "Observium Network Observation System" <observium@observium.org>
Sent: 10/22/2014 4:49:53 AM
Subject: Re: [Observium] 95th percentile and billing module - possible discrepancies
 
Hi Tom

Thanks for getting back to me (and sorry for my slow response!). It is exactly as you say and after chatting with them in more detail cleared up a lot of how they make their calculations. They use RTG and its 95.pl perl script to get the values out -- which adds the two up separately as unrelated links. So it just comes down to two different methods of calculation.

I would like to ask if it were possible to have Observium calculate both the fair way or the unfair way, but I can understand why this may get skipped over (because.. it is unfair!). Tempted to have a look at the calculation code to see if I could help out in anyway by seeing how hard it would be to add it in myself and pass back for everyone to look at?

Thanks again for the info!

Regards

Colin Barker



From:        Tom Laermans <tom.laermans@powersource.cx>
To:        Observium Network Observation System <observium@observium.org>
Date:        15/10/2014 15:30
Subject:        Re: [Observium] 95th percentile and billing module - possible        discrepancies
Sent by:        "observium" <observium-bounces@observium.org>





Hi Colin,

As far as I understand it, if you have 2 ports in a 95% bill, the traffic data will be added together, then 95% is calculated. This is pretty reasonable seeing as that is the data you have actually used and bursted during that time, in total.

If you look at both ports separately, they go higher because you just add up 2 95% numbers that are actually "unrelated", you get 80.

In short, your bill is not fair, but I believe many providers will count it like that... (one of my upstreams does this too)

Tom

On 10/15/2014 04:22 PM,
colin.barker@bluesky.co.uk wrote:
Hi

I thought I would post here first, just in case what I am seeing isn't a bug, and just me just mis-understanding the calculations.


Just a quick setup - we currently are billed on a 95th percentile from an ISP, which we have two physical ethernet connections plugged into two different HP ProCurve switches. I have created a Bill under Traffic Accounting, and set the billing type as 95th @ 50 Mbps, and added both the ports for our connections so I am able to see both under "Bill Ports". This has been running for about 6 - 7 months now, calculating away.


While going back through our reports, I noticed that in September we had an overage marked in Observium as 11.4Mbps (Actual usage 61.4Mbps). However the bill we got from the ISP gave us an overage of 30.38Mbps (Actual usage 80.38Mbps). So I went to look deeper and got the break downs I needed.


Our ISP gave us the following for the highest 95th:


Link1: 50.375Mbps

Link2: 30.005Mbps

Total: 80.38Mbps


I then looked on Observium for the graphs for the two ports, and they reported the following for the highest 95th:


Link1: 49.64Mbps

Link2: 30.44Mbps

Total: 80.08Mbps


I can accept that two different systems there may be a very small discrepancy with the larger numbers - which is fine, however -- the problem arrives when the billing calculations comes in, and shows the values as above (11.4Mbps over, 61.4Mbps usage). I have uploaded the two port graphs and a screen grab from the billing page to show in more detail the above numbers here:
http://imgur.com/a/iErfo

I just want to check I have not missed something that is staring me in the face in regards to the calculations, and if I am, please let me know! Thanks in advance.


Regards


Colin Barker



_______________________________________________
observium mailing list
observium@observium.org
http://postman.memetic.org/cgi-bin/mailman/listinfo/observium

_______________________________________________
observium mailing list
observium@observium.org
http://postman.memetic.org/cgi-bin/mailman/listinfo/observium