[Freeswitch-users] FreeSWITCH in virtual environments

Bote Man bote_radio at botecomm.com
Mon Feb 1 19:26:28 MSK 2016


It seems to me that the XML dialplan routing process would consume more cpu than a short and simple curl request to a database server, no?

 

Anyway, there are probably many things that can be done to streamline performance for a particular application. I know on Windoze it was once suggested to me a long time ago to disable the operating system updates of file modification time stamps, for example. That can add up when a busy system is writing to log files and creating individual CDR files to be read and deleted by an accounting process. I don’t know how this would apply to linux, but it demonstrates how far out of the box one can look for performance improvements.

 

In any case, when discussing virtualization performance it is essential to provide specifics of the instance that runs FreeSWITCH. If someone reports that FS ran very poorly, but does not say that it was a tiny instance that was starved for resources then we can’t evaluate that report fairly. 

 

So the specifics in the report by Volodymyr are useful. Please keep them coming! I am compiling these anecdotes on Confluence for others to read in the future.

 

Thanks.

 

Bote

 

 

From: E. Schmidbauer
Sent: Monday, 01 February, 2016 10:38
Subject: Re: [Freeswitch-users] FreeSWITCH in virtual environments

 

Hi Volodymyr,

We are using xml-curl (i dont think that should have too much affect) and hash counters only.

No database connections, etc... 

Thanks-- 24 threads and 8gb seems like a good start!

Emmanuel

 

On Mon, Feb 1, 2016 at 10:34 AM, Volodymyr Fedorov <lexxua at gmail.com> wrote:

Hello, it depends.

In my case 24 threads and 8gb of ram was quite enough. But I used only xml-dialplan and hash counters.

 

On Mon, Feb 1, 2016 at 4:16 PM, E. Schmidbauer <eschmidbauer at gmail.com> wrote:

Can anyone share the VM memory/cpu specs used in these cases?
We want to run around 300 CPS on FS (running on vmware).

very little transcoding (if any), audio only

How much memory/cpu should be provisioned? 
I see Grant mentioned "single VM host (6 Cores, 12 Threads)" but how much memory?

Thanks,

E

 

On Fri, Jan 29, 2016 at 10:00 PM, servtelar <servtelar at gmail.com> wrote:

Thanks a lot guys for sharing this info. It’s really helpful.

 

 

On Jan 28, 2016, at 6:18 PM, Sergey Safarov <s.safarov at gmail.com> wrote:

 

We have to core ESXi vm with 140 session (70 calls) with have 70 CPU load.

 

Sergey

 

On Fri, Jan 29, 2016 at 2:01 AM, servtelar <servtelar at gmail.com> wrote:

Hi Chad

 

How many legs you are handling with 20 cores on a conference?

 

Regards

 

Gustavo

 

On Jan 28, 2016, at 7:55 PM, Chad Phillips <chad at apartmentlines.com> wrote:

 

I've had very good luck running the newer video branch code on ProfitBricks: https://www.profitbricks.com/

 

As far as I understand, the CPU cycles are guaranteed on their platform. I've had to put as many as 20 cores on a server to handle some of our busier video conference calls, but with that it runs quite smoothly.

 

On Thu, Jan 28, 2016 at 2:15 PM, Dan Edwards <DEdwards at vertical.com> wrote:

I am reviewing the Confluence Virtualization page and had some questions, in particular about VMWare. My company distributes some of its software as a VMWare image file and we were looking to distribute a new product using FS in the same manner. The products operate at a customer premise, on their VMWare infrastructure, not in a cloud environment. Since our customers already have VMWare, switching to a different VM infrastructure is going to hurt, so I am looking for options/alternatives.

First, does anybody know if the virtual timing issues with VMWare have improved since this page was last updated in 2014? Is VMWare still not good enough? Is it possible to throw CPU & memory at this and make VMWare good enough, or is the virtual timing just not workable?

On the virtualization page, there was a comment from 2010 that you might be happy with a High CPU Medium instance on AWS EC2. Certainly workload is a factor here, but I am trying to get my head around how big a machine to perform how small a workload. Is there a place where people talk about their experiences?

Are there other VM platforms that might be acceptable?

Any help or comment is appreciated.

Thank you,
Dan


_________________________________________________________________________



 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.freeswitch.org/pipermail/freeswitch-users/attachments/20160201/56db9f05/attachment.html 


Join us at ClueCon 2016 Aug 8-12, 2016
More information about the FreeSWITCH-users mailing list