You don't need one. The trial will start, and request a registration key. If you don't have one, just press 'Run Trial'.
1) In V4.03 and higher, printing is supported directly.
2) In earlier versions, there is no print functionality. However, the results can be printed (e.g.. for inclusion into a report) as follows.
Method 1: Print via a screen capture
Method 2: Use DLL and your favorite Application
Yes. The following have been confirmed to work perfectly with Windows 2000. You may need to be logged in as Administrator when you enter the Windows program registration key, as the registry is modified.
Can I use the Erlang software to support combined queues (eg. phone + chat + email, ...) supported by common agents?
No. This is a special case. You really have a number of separate queues....one for voice; one for chat; one for email; ... This is NOT directly supported by the standard Erlang formulas, and not supported by this Erlang software. I have checked my most 'heavy' text book on queuing theory...and this case is not discussed.
So......you will have to estimate. I would consider SEPARATE SYSTEMS - one for voice calls; one for chats; etc.. Then estimate the number of required agents to service each queue. Then add these agent numbers - knowing you have OVER-ESTIMATED the real number of required agents.
Yes. See DLLs.
This limit is fixed in Versions 4.x.
See the Equations page.
1) In V4.03 and above, there is a built-in Call Center tool.
2) In Earlier versions; A call center is modeled using either the Erlang C or Delay model. If caller population >= (20 x attendant population)...you can assume 'Infinite' sources, and hence use the Erlang C model. (Otherwise...use the Delay model).
Lets work an example.Lets say......
Can I call the DLL functions from:
Will any of these produce the number of trunks needed to support peak volumes?
The answer is very much YES. The way to use these (any) erlang product is as follows:
1) Calculate your PEAK traffic load offered to the system. This is often called the 'Busy-Hour' traffic. This would be:
(Calls per Hour - in BUSY HOUR) * (Minutes Occupied by
2) Decide what your system will do with 'blocked' calls:
a) Drop them, or
If (a) decide what percentage of this PEAK offered traffic you are prepared to 'loose' - ie. the caller gets a 'busy' or goes into a queue. This might be "1%" or "0.01%" or .... 0.01% or 0.05% are common numbers for telephony.
If (b) decide how long you want your callers to wait on the queue.
3) Use the erlang product to calculate the number of trunks (lines) required to service the PEAK load.
If you download the free demo of the Windows program from the WWW site, the included Help information includes some worked examples. The Erlang products on this site support 6 different 'models', including models for 'queues' and 'dropped calls'.
If you have downloaded the free trial version of 4.04, and after installation the program it asks for a key and does not show the "Run Trial" button:
The Erlang-G program is for modeling ANY system with REQUESTS, SERVERS, (and QUEUES). It is obviously used for voice traffic, where
REQUESTS = calls
QUEUES = phone queues
But, you can use the Erlang-G program for data as well - specially the 'Erlang-C' or 'Delay' models.
REQUESTS = data packets
QUEUES = buffers built into things like routers or TCP/IP stacks on PCs
Do the "Call Center" and "Erlang C" models support 30 minute intervals or only 60 minute intervals ?
Yes. These models ask for a CALL RATE expressed in units of 'Calls per hour',
but that does not mean this call rate must be maintained for exactly one
Should I use Erlang-B or Poisson models?
1) Decide if your source population is
2) Decide what happens to calls that cannot get 'service' (e.g. talk) immediately. Do they
3) Choose model:
4) Use a tool (such as our Erlang-G program) to decide how many voice trunks you need. At this point - it is irrelevant if they are VoIP or non-VoIP trunks.
5) Take the result from step (4) and use the VoIP too to calculate the required bandwidth for your VoIP network.
Generally - Erlang-B is often used in VoIP calculations for infinite populations (or Engset for finite populations) BECAUSE there is often an overflow path (to PSTN?) so loosing a call is not critical; just a little more expensive.
But if there is no overflow path, then the Poisson model (infinite populations) or the Binomial model (finite populations) should be used.
Send mail to email@example.com with
questions or comments about this web site.