erlang B erlang C

rookie1010

Senior member
Mar 7, 2004
984
0
0
Hello

I was wondering if erlang B in general applies to circuit switch traffic &
Erlang C applied to packet switch traffic?
 

orangat

Golden Member
Jun 7, 2004
1,579
0
0
Erlang B applies circuit switched traffic since the formular is the relationship between blocking, trunks and bh.
Erlang C however is applied in both packet and circuit switched traffic.
 

rookie1010

Senior member
Mar 7, 2004
984
0
0
thanks for the reply orangat

how is erlang C applicable to circuit switched traffic, is the additional parameter the length of the call?

are there any other models which are applicable to packet switch?

is Best Effort Packet switched traffic modelled through erlang C?

 

orangat

Golden Member
Jun 7, 2004
1,579
0
0
I'm not too clued in on erlang c but I know its similar to B except that blocking is assumed to be on hold, thus based on queuing theory instead of dropped like erlang b. This queuing formular is used in voice traffic apps for call center calculations and trunks with queuing capability.
Packet switched/internet type traffic is really not suitable to bh modelling. My knowledge is limited there.

 

rookie1010

Senior member
Mar 7, 2004
984
0
0
i guess if i get a delay when trying to make a voice call and then get connected (usually long distance calls trying to connect trunk to trunk) or get dropped after some time is an example of trunks enabling queing and hence erlang C, correct?

any ideas for how to model packet switch traffic

Hey does best effort traffic mean there is zero QoS associated with it and basically if there is a pipe(trunk/server) with say a capacity of 4 Mbps, and there are a bunch of users who require 500 kbps each, then for a best effort traffic calculaton, i just divide 4 Mbps/500 kbps to get the number of users per trunk/server = 8 users to the trunk/server, correct?
 

orangat

Golden Member
Jun 7, 2004
1,579
0
0
No, I don't think so.
Statistical probability of packet loss with std devs would be a better method of modeling.
I think a QOS of 1.0 is the ideal not zero. I'm not sure what you mean in your example. Theres software erlang calculators if you need some examples. Trying to apply erlang c in your example, you would need to find out traffic load say 100 users*500kbps/4000kbps = rounded up some whole number. Then figure how much delay you are willing to stomach say 50ms, then calculate transmit time per packet and plug the whole mess into the calculator.