[HECnet] Revisiting circuit costs

[Johnny Billquist]

Sigh. I hate it when I update myself over and over...

I feel that this scale gave me a little too little manouver room, so I'm 
going to change it slightly.

I'm going to try to stick with:
0-40ms = 3
40-80ms = 4
80-120ms = 5
120-160ms = 6
160-200ms = 7

And I'll see if this works reasonably well. In the interest of 
experimentation, it could also be interesting if others adopted 
something similar. If they think we should use some other scale, I'm 
open for suggestions.

And this would also imply that ethernet would get (in general) a cost of 
3, although if most of your ethernet traffic is going over the bridge, 
you might want to put a higher cost on the circuit.

	Johnny

On 2016-01-20 23:29, Johnny Billquist wrote:
> Since Bob suggested using ping times, and I actually think that is a
> pretty reasonable idea, I sortof have mapped costs this way now.
>
> 0-40ms = 1
> 40-80ms = 2
> 80-120ms = 3
> 120-160ms = 4
> 160-200ms = 5
>
> I do do some rounding of times and values down when close to a lower value.
>
> But this is a suggestion that would maybe give us a somewhat coherent
> cost scale for HECnet.
>
> And for most people, this is mostly relevant for area routing. What you
> then use for cost calculations inside your area do not at any time
> affect anyone outside your area.
>
>      Johnny
>
> On 2016-01-20 20:14, Johnny Billquist wrote:
>> On 2016-01-20 19:38, Steve Davidson wrote:
>>> SG1 to MIM is a consistent 130ms.
>>
>> Thanks. So my idea on 1-2 for some Europe-links, 3 to the east coast,
>> and 5 to the west seems to also map rather well with the ping times...
>>
>> And to make it clear, my links so far as the ethernet, which I mentioned
>> before, I have a rather high level two routing cost on, and one link
>> each to Bob, Steve and Rok.
>>
>> Or, to show it as my tool does:
>> .mnc list
>> Circuit State Cost   Proto   Port  Remote destination
>> IP-0-0  On    5/5    UDP     700
>> 108-65-195-50.lightspeed.sntcca.sbcg:700
>> IP-0-1  On    3/3    UDP     700   eagle.hecnet.org:700
>> IP-0-2  On    2/2    UDP     700   193.2.191.212:700
>> IP-0-3  Off   1/1    Unknown 700   Default:0
>> IP-0-4  Off   1/1    Unknown 700   Default:0
>> IP-0-5  Off   1/1    Unknown 700   Default:0
>> IP-0-6  Off   1/1    Unknown 700   Default:0
>> IP-0-7  Off   1/1    Unknown 700   Default:0
>> .
>>
>> But, this is as the world looks at from MIM. Others need to figure out
>> what makes sense for them, but this can at least give some suggestions
>> on costs.
>>
>>      Johnny
>>
>>>
>>> -Steve
>>>
>>> Sent from my iPhone
>>>
>>>> On Jan 20, 2016, at 10:30, Johnny Billquist <bqt at softjar.se> wrote:
>>>>
>>>>> On 2016-01-20 18:27, Johnny Billquist wrote:
>>>>> On 2016-01-20 18:24, Robert Armstrong wrote:
>>>>>>> However, decnet.jfcl.com do not respond to pings... :-(
>>>>>>
>>>>>>    Sorry, configuration error.  Try again.
>>>>>
>>>>> Great. A very steady 200ms from MIM (resulution is only 20ms on the
>>>>> machine).
>>>>>
>>>>> Now, if we can have a couple of others, to form some idea of a good
>>>>> basis for costs...
>>>>
>>>> One more datapoint: Me in switzerland to MIM is about 60ms.
>>>>
>>>>     Johnny
>>>>
>>>> --
>>>> Johnny Billquist                  || "I'm on a bus
>>>>                                   ||  on a psychedelic trip
>>>> email: bqt at softjar.se             ||  Reading murder books
>>>> pdp is alive!                     ||  tryin' to stay hip" - B. Idol
>>
>>
>
>


-- 
Johnny Billquist                  || "I'm on a bus
                                   ||  on a psychedelic trip
email: bqt at softjar.se             ||  Reading murder books
pdp is alive!                     ||  tryin' to stay hip" - B. Idol