[atlas] Related to Rip Atlas Credit System

Hi Roya,

I cannot reproduce your values

10*3*(int(2040/1500)+1) equals 60 but not 90.

One-off measurement result costs twice more than recurring one.
In this case you get 120.

I just found that it is not covered in the docs. Sorry and thanks for 
pointing!

Victor Naumov

On 4/14/14, 11:48 AM, Roya Ensafi wrote:
> Hi all,
>
> Considering RIPE Atlas credit system,  ping and tracerout unit cost
> calculations are tricky. It closely depends on number of packets in
> train and size of the packets. Based on [1], predicted cost of a
> traceroute measurement with sent packet size of 2040 and 3 as number
> of packets (detail arguments are bellow) should be
> 10*3*(int(2040/1500)+1) = 90 (refer to [1]). However, predicted cost
> calculated by RIPE server is 120. After looking at available
> JavaScript, we couldn't find the predicted cost calculation. It sounds
> like calculation happens on server side and then the the value is
> returned. Is it possible to check the server code i.e. is the cost
> function calculation on server side publicly available?
>
> Thanks,
> Roya Enasfi
>
> [1]: https://atlas.ripe.net/docs/credits
>
> Details for a traceroute measurement : (Per result: 120.0 credits,
> Estimated total cost: 120 credits)
> Arguments: (Interval: 900 seconds, Resolve on probe: No, First hop: 1,
> #packets: 3, Don't fragment: True, Max hops: 32, Tcp: True,Paris: 16,
> Timeout resp.(ms): 4000, Port: 80, Size: 2040)
>