shrinking the line pool to relieve congestion 3

[johnathcc]

Maybe it's not as crazy as it sounds. The situation is: A medical practice) has a T1 with PRI & DID. Most of the time that's enough. But Monday and Tuesday mornings their billing department and their medical assistants start calling out, and patients call in, and the T1 gets saturated.

One solution, of course, is to add a T1. But what about reducing the number of lines available for calling out? If we did that, patients would have a better chance (on average) of successfully calling in. The customer's staff would have more failed outbound calls, but that might not be as expensive as another T1.

Can anyone point me to the formulas for figuring out the right number of lines out versus lines in depending on expected loads? This isn't quite like the usual Erlang calculation (I think) because we're not allocating full-duplex trunks; we're imposing a kind of rationing that improves the odds for inbound callers.

 

[donb01]

I can give you one perspective coming from an environment of a healthcare network with a main hospital with about 1100 devices and 6 connected clinics, etc good for maybe another 1000.

Medical people are WHINY! They will have a hissy fit because their calls are not going out, citing patient care issues and upset customers because they cannot return those phone calls, they will talk about how they can't do their jobs effectively under those conditions, safety issues if they go to grab a line for a 911 call and it isn't there, and the list will go on and on and on... Management will eventually get sick of the noise and put the squeeze on some vendor or other to fix the problem.

You could add some copper POTS lines and use them for outbound calls to relieve the pressure, depending on the hardware configuration of their phone system, but there is a break-even point there too over another PRI, and the customers have a tendency to grab the caller-id off those outbound calls and use that to call you back rather than the numbers they are supposed to - especially in a larger organization where they think they are going around the system. You can have the caller-id blocked, but then some customers with Privacy Manager will not be able to get your calls... it's really a challenge keeping everyone happy and keeping the costs down at the same time.

I would start by having the carrier do a busy study on the PRI if you haven't already. That will give you a solid number to use when determining exactly how short you're falling; then try to determine how much outbound call blocking is happening. If you have enough of both you might be able to justify the second T1 by the amount of potential business they are losing and how much it will improve customer and employee satisfaction all around.

So there's one perspective on it

 

[Delster]

The reply above is fairly good though external calls that get busy won't show on the carriers report as they don't get that far. Your switch bounces them.

You should be able to do some form of call monitoring on your PBX though which will show you trunk usage with busy times etc

This combined with the carrier report should give you some data to convince them to get more trunks in.

Depending on your PBX you might be able to add direct exchange lines to the switch rather than putting them onto POTs thus increasing the total pool of trunks available at all times for all calls. Make outbound calls select the copper lines first leaving the Pri for inbound calls.

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
umm,it IS supposed to do that, right??

 

[SYQUEST]

What kind of PBX is being used here?

....JIM....

 

[johnathcc]

There's no customer PBX. The customer has a KSU, a Nortel MICS ver 7.0 with 1 digital interface (=T1/PRI) card. The phone (& data) service provider is XO. The PRI protocol is NI-2 and all the settings are vanilla.

The customer's probably just going to add another T1, but I'm just wondering if the need to do so has really been demonstrated. The provider gave us a bunch of stats but they're for the other hi-caps, not the PRI line.

 

[SYQUEST]

Since XO is not a real phone company, they probably don't know what a line/busy study is, or the info it can provide. But, then again it only shows incoming info. For outgoing info you might want to post in the NORSTAR forum. I do not know if the NORSTAR has traffic statistic capabilities, that you would find on the Meridian 1 stuff.

Because the NORSTAR is considered a hybrid system, it can be called a PBX!

....JIM....

 

[Delster]

Meridian does have traffic reporting if thats whats in there. Don't ask me how though as i've very little experience on nortel kit....horrible stuff to work with!

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
umm,it IS supposed to do that, right??

 

[SYQUEST]

All the years I worked on the NT Meridian, it was never horrible. You just need to do your homework, learn and understand the system.

....JIM....

 

[Delster]

The would depend on what your comparing it too! Much prefer ericsson and avaya kit.

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
umm,it IS supposed to do that, right??

 

[donb01]

It sounds kind of lame, but even though I can go through the hassle of internal traffic metering for my outbound traffic it's much easier to tell if there is a call blocking problem by whether or not I'm getting service tickets from people saying when they try to place an outbound call they get a fast busy, or they dial 9 and don't get a second dial tone, or things along those lines. People in my organization will start in on me right away - THEN I can go through the effort of figuring out what's blocking the outbound calls!

 

[DBrewsky]

On a Norstar you can manage the number of trunks for inbound and outbound. For example: On a PRI, you can limit outbound to only use 6 trunks, and reserve the other 17 for inbound. This can be made on the fly. So if on Mondays you wanted to increase the inbound traffic, you can increase the channels as needed, up to 23 of course.

This is done in the CBC Limits. If these values are set, default is 23 inbound and 23 outbound, you can also view the number of times a call was rejected when those maximums were hit. I know when setting up two PRI's, you would generally increase both options under PRI-A to use the additional trunk capabilities.

--DB

 

[johnathcc]

Thanks all. Just for curiosity's sake, the outcome was: the customer decided that it would be more pro-active to just add another line, which we completed last week. I agreed with their decision, but I wanted to make sure I'd explored the alternative thoroughly.

 

[KE407122]

Normal Traffic Engineering -rule of thumb is to have a
1 % chance of Calls blocked on incoming and a 5% chance of blocking on outgoing. In most cases, if you engineer to the Average Busy Hour you use a 1% chance of blocking. (In many businesses the peaks times occur around 10:30 to 10:40 am and 2:30 to 3:40 pm.)


Basics of Traffic Engineering
See Poisson Table (Trunk capacity tables)
See Alternate Routing Table

Both use CCS or Hundred Call Seconds for use in probability of blocking and offered calls. (1 CCS = 100 seconds of telephone traffic.)

KE407122
'Who is this guy named Lo Cel and why does he keep paging me?'

 

[KE407122]

My favorite for changing incoming call capacity on the fly was the old Break-Hunt Switch that was able to toggle to reduce or increase the number of hunting lines on an equivalent group.
Old Old old old ol o ...what was I talking about?







KE407122
'Who is this guy named Lo Cel and why does he keep paging me?'