<br><br><div class="gmail_quote">2012/7/30 Kannan <span dir="ltr"><<a href="mailto:vasdeveloper@gmail.com" target="_blank">vasdeveloper@gmail.com</a>></span><br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
Hi<div><br></div><div>I came across couple of pointers on the Internet regarding solutions available for providing hosted PBX service.</div><div><br></div><div>1. Multiple PBXs: Using separate hardware to host each PBX. Pretty straightforward, but no hosting company wants to use it.</div>
<div>2. Multi-tenant PBX: Configuring multiple PBXs within the same instance of Asterisk. I.e. partitioning a single instance of Asterisk into multiple PBXs by way of configurations, using unique landing context for each tenant.</div>
<div>3. Virtual PBX: Multiple virtual machines within the same hardware, each host an instance of Asterisk.</div><div><br></div><div>Which one of the method above is generally used by hosted PBX service providers?</div><div>
<br></div><div>Isn't the second option with ARA a good choice for dynamic creation of multiple "small" PBX tenants?</div><div><br></div><div>Is the last option alone or combination of options 2 and 3 good for cloud based hosted PBX service offering?</div>
<div><br></div><div>Thanks,</div><div>Kannan.</div>
</blockquote></div><br><div>Working in the voip field from a lots of years, I have found all three type of business.</div><div><br></div><div>The first is maybe the easier and most common. Hardware is cheap and it is easier to "sell" a service like the PBX if it is sold together with a piece of iron. Usually the hardware is placed on client's network, using the bandwidth of the client. Usually together with the PBX is sold also a router/firewall/traffic shaper/vpn endpoint to try to optimize the traffic on the client's DSL.</div>
<div><br></div><div>The major pros about this solution is you can use a normal PBX like freepbx/trixbox, the client can mess the config how he likes, without disrupting other services, you can install VoIP card to connect landlines,.</div>
<div><br></div><div>The major cons is the cost of the hardware, the cost of the g.729 licenses (if any) and the maintenance cost of replacing hardware failures and the need to be physically near each client.</div><div><br>
</div><div>The second is the holy grail of the VoIP providers. </div><div><br></div><div>The major pros is the cost. Having a single hardware is cheap and it is still cheap also if you decide to get two to be ready in case of an hardware failure. </div>
<div><br></div><div>The major cons is the software. You cannot use the award winning freepbx/trixbox family and you need to deal with sometime limited or incomplete developed interfaces. The client always asks for the missing feature. One other major cons is the "reload". If the PBX software is not made using ARA, then every time you add a new peer or a new DID, you need to reload the entire PBX and that is a resource killer. Again, if the pbx interface is not made using ARA, then you cannot let your clients to change the configuration or they will trigger continuous reload (and delaying reload for example every 10 minutes is not a solution)</div>
<div><br></div><div>The last one is sometime the chosen compromise, but from my point of view, pbxes are not good software to virtualize. They are too sensible to delays and your voice quality can go down if the real server is overloaded.</div>
<div><br></div><div>The same for the cloud based solutions (I have yet to found). I suspect the "cloud" is good for services like http, not for real time applications.</div><div><br></div><div>Leandro</div><div>
<br></div>