Hi everyone!
In my company we started deploying the IP telephony system in the Cloud chosing Amazon as provider. We want a system fault tolerant with high availability and my question is if anyone has implemented this already. If so, can he/she give me some description on how he/she did it.
Thanks in advance!
If you want high availability, you need to control the hardware.
I’d be surprised if any cloud implementation was immune from choppy audio.
Well, virtually speaking, you have control over the hardware as Amazon guarantees a minimum performance and bandwidth.
What’s the averaging time for their performance measurements. VoIP needs an averaging time of < 20ms.
I don’t understand what you mean. Amazon measures the process power with EC2 units (one EC2 unit equals to the CPU capacity given by a Opteron 2007 or Xeon 2007 1,0-1,2 GHz processor). From there you can set your instance accordingly to your needs.
We have been using AWS for 2 months already with perfect results. Now we want to set High Availability. Cloud performance is not the issue here unless it has anything to do with high availability environment.
If you average over one second, a 5GHz single core can easily give you 1GHz effective speed even though you only have access to it for the first 250ms in each second. That would be useless for VoIP, as you would end up with 750ms+ latencies and corresponding peak to peak jitter.
To avoid significantly increasing jitter, you need that 1GHz share averaged over every single 20ms interval, and preferably over every single 5ms interval.
Hi using asterisk on a AWS instance is ok, we have customers doing just this.
But what we found was that teh cost was similar to using a hosted server for which we have full control over.
not looked at teh high avaliblity aspect of it. as my vps supplier now supplies redundant VPS’s