Setting up VoIP provider network

Hi,

I am looking into starting my VoIP business by setting up VoIP provider network in 4-5 different geographical locations.

I have some questions:
1- What hardware I would need for IP-PSTN gateways?
2- What bandwidth I should expect.
3- How much should I expect such a network to cost me?

Thanks!

hi in brief,
1. check the possibility to peer with other providers or terminator that will accept SIP trunk. it will save you huge a mount, today and in the future.

2.it depend on your design infrastructure. would you pass all your calls on your SBC or not...? this will make a different. other part is your endpoint in use are they Nat traversal support or not, this will force the transport to go thru the SBC.

3.you can do it with 1K pc and up to 500K$ and more. its all depends how many subscribers you want to support and how you want to do it... you need to spend on top of the "voice system" more on CRM, Billing, FW and redundancy.

good luck

ab1canoby :

Hi, I am looking into starting my VoIP business by setting up VoIP provider network in 4-5 different geographical locations. I have some questions: 1- What hardware I would need for IP-PSTN gateways? 2- What bandwidth I should expect. 3- How much should I expect such a network to cost me? Thanks!

i am not an expert but from my basic knowledge i would say the following: you could certainly go into business (at least in most western or developed countries) without any of your own hardware if you wanted to use hosted solutions for your billing platform, etc. and and SIP trunks for all the interconnect to the PSTN(therefore not needing any of your own gateways) and you could be global without having to install equipment in different locations. i would suspect that your biggest up front investments would be getting your billing systems up and running and prepayment of wholesale termination capacity(you may be able to get credit; but it is my understanding most startups have to pay in advance unless they are very highly financiered)

as far as bandwidth if you will be connecting through your own gateways you would need in the range of 40kbps to 80 kbps depending on the QoS you are trying to achieve and the codec necessary for the other components of your network.

i believe that your cost will be very dependent on the geographic locations that you set up. defiantly your IP link for example would be only a fraction in western Europe to what you would pay for a similar connection say in Africa. you can start a VOIP business with anything from a couple hundred dollars to many millions. i believe that the startups with the multi million dollar investment the big bulk tends to go to marketing efforts.

if you are thinking along the lines of starting a calling card or discount calling type company you may want look at options such as flatplanetphone. there you can start your own brand for less than $200 and do not not need any equipment or IP capacity. the trade off is that you will not be able to offer really low prices that you can offer if you deal direct with wholesale termination providers and route through your own switches. anyways you can even set up a trial account for no investment at all and they give you $15 credit just so that you can see how it all works.

one other thing to add. if you will be using your own gateways i would closely investigate the regulatory status in the countries you are interested in setting up your switches. there can be some very serious legal hurdles to go through that could cost quite a bit more that the hardware or IP capacity to get started. many countries such as india, most of the middle east, and most of africa are pretty much off limits territory for VOIP at the moment at least if you were thinking of installing your own switches there.

just out of curiosity; what 4 or 5 locations are you thinking about?

spg

Thanks DesecondG for your response!

I plan to privde VoIP services for ~ 40 Call Shops from one georaphic location to ~5 other locations.

Each Call Shop has an average of 12 IP phones, expecting ~450 simultaneous users at peak time.

Currently, the local shops use mixed provides (net2phone, etc) for the lowest rates to most frequently called geographies.

I would need to better estimate my initial investment to know how long it would take me to recover my expenses.

I am surprised that peer terminators would cost me less, I thought my running expenses would be to a minimum if I setup my own network, could you clarify or point me to more resource on peer termination?

Thanks!

Quote:

I plan to privde VoIP services for ~ 40 Call Shops from one georaphic location to ~5 other locations. Each Call Shop has an average of 12 IP phones, expecting ~450 simultaneous users at peak time.

The only way to be sure that you are going about this in the right manner would be to decide exactly where your prospective customer wants the balance between quality and cost.

If those customers want quality then building your own network is one way of ensuring that. The cost will inevitably be higher and that's a cost you have to pass on to the customer. Going that route you also need to decide where you want your control to stop - do you just need an IP network or do you want to own and control the media servers, SBC's and gateways? Again, this all comes down to quality of service versus cost.

If they don't really care too much about quality as long as it's usable, its pointless going to the expense of building a network.

If you're looking for something in the middle I'd be minded to hook up to a white label service with their own gateways and billing engine, and build a simple IP WLAN by buying up some fibre interconnects from the five locations back to the white label providers datacentre, and one interconnect to a decent datacentre in the location of the callshops.

You'd need around a 20-30mb pipe for which I would I have previously been quoted £1,000/month based on a US/London interconnect. You'd need to check each of those countries for pricing. You'd also need a couple of U rack space at either end of each.

You'll get an overview on building VoIP networks and costings in this thread (although given you're only servicing callshops much of the security/SBC/NAT issues can probably be avoided):-

http://www.voipuser.org/forum_topic_8289.html

Dean

ab1canoby,

<snip>

one more interesting thing is Deen point regarding the pipeline, i dont think you need to wire all your callshops wherever they are... the beutty with IP is that they can register directly to the SIP provider and you dont need to care more than collect the calls charges.
Deen, am i wrong here?


<edit by Dean>Recommending specific providers just causes us spam issues - thanks</Edit>

Quote:

Deen, am i wrong here?

It's not a case of right or wrong, it's a case of quality of services versus cost and the nature of the customer you're marketing to.

If QoS is an issue, you need a QoS IP backbone and owning your own pipes to countries of termination (especially if it's only 5) is an option.

dean :

If you're looking for something in the middle I'd be minded to hook up to a white label service with their own gateways and billing engine, and build a simple IP WLAN by buying up some fibre interconnects from the five locations back to the white label providers datacentre, and one interconnect to a decent datacentre in the location of the callshops.

Thanks for you suggestions. I am looking for something in the middle between quality and cost.

Why would I need an IP WLAN if I hook up to the white label? Wouldn't the while label service inculde the IP-PSTN termination?

Thanks!

Quote:

Wouldn't the while label service inculde the IP-PSTN termination?

Termination yes. I'm talking about what happens before that. If your customers connect to the PSTN gateway via the internet you have no control over QoS. If they connect via your private LAN, you do.

If all of your traffic is destined for the PSTN, then you only actually need a single pipe from the call shops to the PSTN gateway of your provider.

Let's say for example these call shops were in New York and the white label provider had their PSTN gateway in London. You would buy a (say) 32mb fiber interconnect between New York and the London providers datacentre. At the London end you would extend the white label operators LAN to New York by offering access to their network from New York (the other end of your fiber).

This is just standard IP networking. Have a read on BGP to understand how this works (it's largely automatic):-

http://en.wikipedia.org/wiki/Border_Gateway_Protocol

Once you have that private line, you would have full QoS over the transatlantic connection. So customers would connect over the internet initially, but only for the brief few hops to your private fiber pipe. Using BGP call shops in New York would experience very good QoS as long as the call shop itself has decent internet connectivity and is only a couple of hops from your fiber interconnect. As you moved, say, to the west coast US, your QoS control would drop, but if that became a problem you could just extend your own WLAN to a couple of extra entry points across the US.

I'm obviously over-simplifying this to save having to write the manual on how to build it, but you'll get the gist of it I'm sure.

Dean

dean :

Quote: Wouldn't the while label service inculde the IP-PSTN termination? Termination yes. I'm talking about what happens before that. If your customers connect to the PSTN gateway via the internet you have no control over QoS. If they connect via your private LAN, you do. If all of your traffic is destined for the PSTN, then you only actually need a single pipe from the call shops to the PSTN gateway of your provider. Let's say for example these call shops were in New York and the white label provider had their PSTN gateway in London. You would buy a (say) 32mb fiber interconnect between New York and the London providers datacentre. At the London end you would extend the white label operators LAN to New York by offering access to their network from New York (the other end of your fiber). This is just standard IP networking. Have a read on BGP to understand how this works (it's largely automatic):- http://en.wikipedia.org/wiki/Border_Gateway_Protocol Once you have that private line, you would have full QoS over the transatlantic connection. So customers would connect over the internet initially, but only for the brief few hops to your private fiber pipe. Using BGP call shops in New York would experience very good QoS as long as the call shop itself has decent internet connectivity and is only a couple of hops from your fiber interconnect. As you moved, say, to the west coast US, your QoS control would drop, but if that became a problem you could just extend your own WLAN to a couple of extra entry points across the US. I'm obviously over-simplifying this to save having to write the manual on how to build it, but you'll get the gist of it I'm sure. Dean

Dean,

i am wondering if the termination partner in london was a low cost operator (which is what i tend to believe most callshops use) would it not be expected that the calls will be leaving that partner via the public internet; particularly if the calls were being billed at discount rates to third world countries(which i believe is where most callshop calls terminate)? if this is the case would there not be very little QoS advantage to use private fiber from new york to london (agruably one of the most reliable IP routes for public internet)? after all the primary link for call degradation would tend to be that final hop to africa, east asia, etc. or at least that has always been what i have thought when it comes to international termination.

my instinct(and i am really not an expert - or even close) and basic knowledge would tell me that for a call shop interested in offering the best prices(aren't they all) may be making a better investment in establishing TDM termination directly to the most called destinations than leasing fiber across the atlantic. the way i see it you could route calls for that first hop(NY to London) via the pstn if needed for quality and likely still spend less than a leased fiber connect every month; but owning that final connection to a place like Ethiopia or Congo(or any destination popular with the call shop crowd - every call shop i have seen is usually full of migrant immigrant workers calling home) is where you may be able to make significant savings over white label wholesale rates.

spg

p.s. please do not take this as any kind of expert opinion; i am writing this as much to learn(by listening to replies) more about the business myself as any other reason.

Quote:

basic knowledge would tell me that for a call shop interested in offering the best prices

My knowledge tells me the same but experience tells me to ask the question, and the answer was actually not best price centric:-

"I am looking for something in the middle between quality and cost."

Quote:

if the termination partner in london was a low cost operator (which is what i tend to believe most callshops use) would it not be expected that the calls will be leaving that partner via the public internet

Again, that's not the case here if the original poster answered the QoS/Cost question accurately.

Quote:

one of the most reliable IP routes for public internet

New York was of course just an example assuming the call shops were in the USA. I'm not suggesting that all call-shop operators should run a dedicated pipe over to 60 Hudson just for the sake of it

Deanw

you find this page to find the right solutions, like other people said , all depends on how many subscribers, bandwidth, etc..

Another advise don't buy equipment but rather try the hosted solution. Many people thinks that with 1 T1 data they could offer a voip service, wrong, wrong!

Quote:

1 T1 data they could offer a voip service, wrong, wrong!

That depends entirely on the requirements for the service. A T1 pipe coupled to a POTS line backup would be fine for many businesses.