Getting Started- Deciding on Asterix


#1

I understand that this is a very technical forum on Asterix and I have read the other intro info on Asterisk. However I have not been able to find impartial, honest (no sales pitch) open source PABX advice on the web. So I thought I would ask these questions here to advanced users.

  1. Firstly do you know of a site that compares alternative open source PABX products?
  2. Do you need to be an IT programmer to integrate a solution or could it be setup by a user that understands PABX’s?
  3. Does Asterisk require ongoing maintenance and at what intervals?
  4. Is there a general amount of time to set up the initial basic configuration (assuming continual tweaking and adjusting after the initial setup) ? eg. Two weeks or one day as suggested in the Asterix@home newbies guide.

Any thoughts would be appreciated as I am considering the learn and build or buy option as I am starting up my new business.


#2
  1. no, nothing that isn’t slanted one way or the other…voip-info is probably the most unbiased, but even there, a large portion of the site is dedicated to asterisk.

  2. you don’t NEED to be a programmer, but if you understand general programming principles and technology in general, you should be fine - just be prepared to do a LOT of reading on voip-info trying to learn all of the cool things you can do.

  3. i wouldn’t say it requires maintenance, but digium is currently shooting for a release every 6 months, so if you wanted the newest features and fixes, you’d need to upgrade to the newest version. this is fairly easy to do and most of the time, you don’t need to modify your dialplan much, so it’s not a difficult proposition. as far as nightly or weekly stuff…we just reboot the servers once a week and don’t have too many issues.

  4. i can do a setup in one night now, but when i started, it would have taken me an entire week…it’s just like any other complex application - you need time to learn. if you’re serious about going with asterisk as a business, forget about *@home - go download the source and build a PBX from that - it’s the only way you’ll learn how to use the software. if you’re just looking for a PBX but won’t be doing anything more with it, then *@home would probably suffice.

i will tell you that we replaced almost $2 million worth of phone system with $50,000 worth of equipment to run asterisk and haven’t even considered looking back. FWIW…

HTH.


#3

looking over what is written above all i can say is ditto. i was not in the telephony business before i started with asterisk and didnt know much. however once i became involved it wasnt very hard to pick things up. a little over a year later i have my own voip company. we are in the process of building multiple systems for others. we have looked at other solutions out there for cost, quality etc. and we are still sticking to asterisk.


#4

Todd,

One of the things I’m having to consider very carefully, is the aspect of a user’s ability and willingness to adapt to the way Asterisk functions when dialing out.

Everyone gets very accustomed to having an instantaneous response to most calls dialed from a standard piece of phone equipment. It’s called “early-dial,” where-in each digit dialed is acted upon immediately by the “Central Office” switching equipment - and when the full number is dialed, the call is immediately processed, ringing the remote destination right away.

This is not the case with most calling situations using Asterisk. With some SIP phones, the process goes much more like using a cellular phone, where one dials the number, then presses the send button (or holds on while the digit time-out period expires).

Zap calls (Asterisk, connected directly to standard analog phone and phone company analog circuits) are even more trying in that respect, as there seems to be no way of getting the dialed call to process until after the digit time-out. In my experience here, where I’m developing a replacement for an aging Mitel analog switch, that period is about 8 seconds of silence. That confuses folks, who often just hangup, thinking nothing is happening, and give it another try. If the hangup isn’t long enough, then they have unwittingly executed some other operation like initiating a three-way, putting the first line on hold, getting music going somewhere, and generally confusing the issue all the more.

I’m hoping that not too long from now a reliable method of instituting “early-dial” will come about.

All that said, I’m a novice and might be blowing more smoke than reliable information - I admit that. But I’m surely running all this stuff through my feeble, smallish mind often.


#5

Brian,

just out of curiosity, what signalling do your trunks use? we’re primarily robbed-bit, but are moving to PRI - the one PRI T we have converted almost instantaneously connects the number once the dial command has been sent…with pattern matching on the polycom phones, it’s ALMOST like a regular phone in terms of speed.

just thought i’d share…perhaps it’s not the server, but the trunk technology you are using…


#6

Hth, and where every thing’s up-to-date

Ya, we’re using 5 pots (ground-start) lines for the Mitel, and have three other standard loop-start pots lines - 1 pay phone, and two versitile lines (fire alarm, fax, my Asterisk ZAP connector, etc.)

Like I said, I’m a novice with digital phone systems, but have a pretty good feel for the old tried and true. Our Mitel is an early '80s vitage switch with about 75 extensions (8 SuperSet’s) right now. It’s tired and so am I of trying to keep it up and running.

I’m sure we will be maintaining the PSTN analog lines for some time. We will also be maintaining quite a number of analog extensions - requiring us to connect a couple of channel banks. That’s where I think things are going to differ significantly for the users that are accustomed to the standard methods.

The other thing that has me scratching my head is the attendant console thing. We have two positions going simultaneously in parallel - one used during regular business hours fielding most daytime calls. The other is a 24 hour position - generally being manned by persons who have little or no PBX system skills and sometimes few other skills. (Don’t tell anyone here I said that :smile:) Although the old Mitel consoles have a lot of buttons, very few of the daily operations require much learning and effort. Transferring calls and parking is going to be quite a bit more complex and requiring several more key presses than the present.

I suppose people adapt, but I’m still occasionally struck with the thought that we’re kinda’ going backward in some respects as far as ease of use of a multi-user phone system goes.

Thanks for allowing me to use this particular discussion thread for my own personal and public thinking.


#7

Thanks very much for the candid feedback. It was also good to get an understanding of people just trialling * at the moment. I would be using POTS analogue lines into the building, have other users experienced the sorts of delays mentioned by Brian using the Analogue lines.


#8

Good morning (USA - Pacific time), Todd,

My experience with this so far has been with a Digium TDM400b card onto which I have one FXO trunk line card and two FXS extension line cards.

I think the best nut-shell explanation of what I’m describing is that all of the dialed digits have to be collected first, then a determination has to be made by Asterisk (in the case of ZAP phone equipment) that the dialing has ended (by a “timeout”), then the trunk is seized and the dial string issued out.

That’s a different process than just picking up a phone and dialing - even through a more conventional PABX, which makes the connection of the extension to the trunk first, then your DTMF tones sent in real time down the line to the CO.

Again, it’s just about like doing a call on the cell phone, except with the cell and/or a SIP extension phone, one can press “send” and get the call going. Not so with a standard phone connected to an FXS port on Asterisk.

I’ve briefly read of some scheme of dialing an access digit to get an outside (ZAP/PSTN) line - just like with the old PABX - but I haven’t really persued it, and I also seem to remember reading about it being kind of kludgy - maybe somemore of that delay stuff.

And finally, I reiterate my non-position of authority and expertness in most of this stuff. I surely invite anyone who knows better to set me and the record straighter.


#9

ah, ok…

i don’t have any experience with the smaller analog cards - we’re only using quad span T1 cards in the systems, and have a minimum of 48 trunks per server. i’m not a telco guy myself, more of a sys admin and programmer/project manager, but we’re talking apples and oranges. you’re using analog, we’re using digital. however, we’re running with an older and slower signalling type, which is why i mentioned the PRI circuits.

i will keep your comments in mind, though, as i move forward, as we may be deploying smaller boxes with analog lines in external locations…if there is that much of a delay, we may have to rethink some of the architecture.

thanks.


#10

[quote=“bingoldsby”]Good morning (USA - Pacific time),
I think the best nut-shell explanation of what I’m describing is that all of the dialed digits have to be collected first, then a determination has to be made by Asterisk (in the case of ZAP phone equipment) that the dialing has ended (by a “timeout”), then the trunk is seized and the dial string issued out.
[/quote]

Let me point out that I’m also an Asterisk newbie, so the following could be wrong. But I’ve stumbled into the same problem as you and have had a veteran here trying to explain to me what is going on. Basically, due to the difference in how circuit switched network (POTS systems) and packet switched networks (VoIP) function, the problem you are describing cannot be solved.

With an analog phone, each digit you enter extends the channel you have reserved between you and an end point. For example, if you press 555 you will have reserved a channel from you to some node on the 555 area code. Then you enter some more digits and your continuation is extended until you reach your callee that is the end point and whos phone is now ringing. You reach your callee by taking a one digit step at a time.

In a packet switched network the above doesn’t work. You don’t know which route your packets will go and you can’t reserve any channel on the network. That is why you first have to tell your phone that you are done entering your number and then have it establish a route to your callee. The situation is similar to entering an URL in a web browser.

I do agree with you 100% though. This is definitely a regression from using a POTS. It can be helped by using phones that has a configurable No Key Entry Timeout setting. With those you can choose how long the delay is before the phone sends the number. Unfortunately, it is not all phones that support setting the No Key Entry Timeout settings. We had a few ones with a too long unconfigurable timeout and they were very frustrating for many users who didn’t understand that they should use the # button to send the number. Eventally, we got some new ones with a saner No Key Entry Timeout setting.


#11

Actually bjourne, that’s not entirely true.

The public switched telephone network, (and traditional PBX’s) work on the “first dialed digit” principal.

If you’re in your PBX, you can place 3, 4, or 5 digit extension calls, because the PBX has been programmed to know that when you dial the first digit in an extension, it should expect to see 2, 3, or 4 more digits. Once it gets the last digit it’s expecting to hear, it starts routing.

If you’re dialing an outside number, you may have to dial an access digit. (9 for example) When you dial that 9, the system is programmed to give you dial tone back as feedback, and then it expects to hear a number of digits corresponding to the number you need for your particular country.

If things worked as you describe, people would be taking up circuits as they dial. Even if they mis-dial, hangup, and re-dial a number. It would be an inefficient use of the network to be sure. It USED to work as you describe, but hasn’t been so since the times when rotary dial ruled the world.

As everyone points out, VOIP calling does not have a destination until dialing is complete. To make those work there are a number of solutions:

The “send” or “call” button can be used to signal end of dialing to the client, which will send the completed number to the VOIP network for processing. This is similar in operation to that of a cellular/mobile phone call.

The client can be programmed with it’s own dial plan information so that certain digit combinations are, by default, considered “complete phone numbers”. When the client hears a complete phone number, it sends it to the VOIP system for processing.

“Look ahead” dialing, (which not all clients support) will send any digits you dial one at a time, adding each new digit, until the VOIP system you’re working with understands the destination. That is, you may be trying to dial extension 303. When you dial the first digit, your phone sends a “I’m calling extension 3” message to the VOIP system. The VOIP system responds, “I don’t know anything about an extension ‘3’.” You then dial your second digit, and the client sends the message “I’m calling extension 30”, which again gets rejected as an incorrect destination. This continues until the VOIP system gets a destination it’s programmed to understand, whereupon call processing proceeds. This system has the disadvantage of requiring the VOIP system to send 8 to 10 reject messages for every message it sends confirming that it is processing the call, but it’s closest to what the traditional telecom network does.

None of these are perfect systems, but that’s because we’re trying to fit the phone number destination system into an IP network design. I’m not a seer or anything but I’m guessing that soon enough, the phone number will disappear, and be replaced by a more normal network system name. (calling: user@somenetwork.net)