me_nortel_guy

According to the NTP (553-3001-151) section regarding how many superloops are required a superloop can support 120 tns and with guarentee of no blocking and up to 1024 if blocking is allowed. Sounds to me like you may be closer to the limit now than you may think.

slagburn

What type of situation are we looking at here... regular old office? Call Center? Railroad emergency phones? All factors to consider when looking at time slots.

I have a couple of 61s with 400 users on 2 IPE mods on 1 superloop and no problems because it's a very low traffic site, on the flip side I have a call center with 2 super loops per IPE mod and a few IPE mods due to high traffic.

bwells

I should've said this before I guess. It is not a call center, about 300 TNs are regular offices which are open during normal hours. The other 950 or so are student dorm rooms which dont get busy till after classes. This mean I usually have about a 2 hour window which the usage patterns overlap. normally fron around 3 to 5, Like I said before so far I dont have any problems with blocked calls, yet. I know there must be some general rule of thumb that can be used to estimate the number of time slots to TN ration.

Fletch

Try running some traffic reports. They do include information other than just trunk usage. You can see CPU utilization, and timeslots as well.

This will give you an idea of where you are today with your current loading, and then help you plan better for the future.

__________________
Fletch

For more on Nortel E911 Solutions on the web:
http://nortel.com/e911

bwells

Thanks Fletch, I hadn't thought of doing that. I'll give it a try.

rhack

When engineering a switch, I try to engineer for a minimum grade of service per station. Depending on the application, different gos’ are used. For low usage applications, (hospital, dorm etc) typically 2.5ccs per station is acceptable, for heavy business use, 7.5ccs is what I typically use. For heavy use ACD or trading type applications you may want to increase the ccs per station. For a mix of business, and low usage, a happy medium of 5 ccs may be something you want to shoot for. So what does all of this mean? If a user is on the telephone for 1 full hour, they would be utilizing 36ccs or 1 timeslot for the entire hour. So 2.5ccs = 7 min, 5ccs=14 min, 7.5ccs=21 min.

So lets say, you jam pack every possible slot on an IPE and use all 256 ports available (I’m not counting the potential for data units). Now assume you want to give each telephone a middle of the road 5ccs. 256 * 5 = 1280ccs handling capacity required. If you have provisioned one superloop you have provided a total carrying capacity of 4320 ccs (120 * 36ccs). Looks like it works with basic math; appears that only 30% of available capacity is needed to provide 5ccs per station (1280ccs / 4320 ccs = 30%). Take it a step further and do an erlang or erlangB calculation to see how many channels or timeslots you need carry the traffic. (see www.erlang.com)

Before using the calculator you need to convert ccs to erlangs. (ccs / 36 = erlangs). So, 1280 / 36 = 36 erlangs. Using the extended erlang B calculator, if you plug in 36 erlangs for your BHT (Busy Hour Traffic), .01 for blocking (1 call in 100 will be blocked), and a 50% retry rate (50 % of blocked attempts will immediately try to re-attempt). The result shows 49 lines (in our case Time Slots) needed. This equates to 41% (49 / 120) of the time slot capacity is required to handle the 5 ccs per station. It is higher than the 30% derived from the straight math, as retries and other statistical reasoning is taken into consideration with the erlang formula. You can tweak the BHT, gos (try .001) and retry attempts to see the impact on the lines required.