VoIP Mailing List Archives
Mailing list archives for the VoIP community |
|
| View previous topic :: View next topic |
| Author |
Message |
rob at hillis.dyndns.org
Guest
|
 Posted: Fri Apr 25, 2008 10:10 am Post subject: [asterisk-users] Quality problems with ISDN PRI |
 |
|
To the best of my knowledge, multi-core processors are not hyperthreaded
- certainly my Core 2 Quad processor isn't. I would expect a Core 2 Duo
to be the same.
Steve Totaro wrote:
| Quote: | On Thu, Apr 24, 2008 at 11:18 AM, Rob Hillis <rob at hillis.dyndns.org> wrote:
| Quote: | Every CPU core shows up as a separate CPU under Linux. For those that have
hyperthreaded processors, a single core processor will show up as two
processors - assuming you have hyperthreading enabled.
|
That is interesting. I have an intel C2D and I can only see two
procs, not four, is that normal? Are you sure what you are saying is
correct? I am obviously running SMP.
Thanks
Steve totaro
_______________________________________________
-- Bandwidth and Colocation Provided by http://www.api-digital.com --
asterisk-users mailing list
To UNSUBSCRIBE or update options visit:
http://lists.digium.com/mailman/listinfo/asterisk-users
!DSPAM:4811f27e213018250212299!
|
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.digium.com/pipermail/asterisk-users/attachments/20080426/4e93fdae/attachment.htm |
|
| Back to top |
|
 |
stotaro at totarotechn...
Guest
|
| Posted: Fri Apr 25, 2008 11:22 am Post subject: [asterisk-users] Quality problems with ISDN PRI |
|
|
On Thu, Apr 24, 2008 at 4:45 PM, Jay R. Ashworth <jra at baylink.com> wrote:
| Quote: | On Wed, Apr 23, 2008 at 02:14:27PM -0400, Steve Totaro wrote:
| Quote: | There are much better solutions than doing a RAM drive. While it may
be stable (not in my experience, I advise using different servers for
different tasks (with redundancy obviously). A phone switch should be
just that, a recording server should also be just that (in demanding
environments).
|
That would be fine, if Asterisk was capable of buffering recording
writes, but I'm told it's not; the I/O involved in getting that
recording data off the box in real time is probably worse than that of
putting it onto disk -- disks are usually higher bandwidth channels
than network adapters.
For permanent storage, certainly, the recordings should be moved to
another box, and that's how we do it here.
Cheers,
-- jr '44 byte chunks. Is someone an ATM fan?' a
--
Jay R. Ashworth Baylink jra at baylink.com
Designer The Things I Think RFC 2100
Ashworth & Associates http://baylink.pitas.com '87 e24
St Petersburg FL USA http://photo.imageinc.us +1 727 647 1274
Those who cast the vote decide nothing.
Those who count the vote decide everything.
-- (Joseph Stalin)
|
Well in the real world, your hypothesis has been proven wrong.
Thanks,
Steve Totaro |
|
| Back to top |
|
|
stotaro at totarotechn...
Guest
|
| Posted: Fri Apr 25, 2008 11:24 am Post subject: [asterisk-users] Quality problems with ISDN PRI |
|
|
On Fri, Apr 25, 2008 at 12:22 PM, Steve Totaro
<stotaro at totarotechnologies.com> wrote:
| Quote: | On Thu, Apr 24, 2008 at 4:45 PM, Jay R. Ashworth <jra at baylink.com> wrote:
| Quote: | On Wed, Apr 23, 2008 at 02:14:27PM -0400, Steve Totaro wrote:
| Quote: | There are much better solutions than doing a RAM drive. While it may
be stable (not in my experience, I advise using different servers for
different tasks (with redundancy obviously). A phone switch should be
just that, a recording server should also be just that (in demanding
environments).
|
That would be fine, if Asterisk was capable of buffering recording
writes, but I'm told it's not; the I/O involved in getting that
recording data off the box in real time is probably worse than that of
putting it onto disk -- disks are usually higher bandwidth channels
than network adapters.
For permanent storage, certainly, the recordings should be moved to
another box, and that's how we do it here.
Cheers,
-- jr '44 byte chunks. Is someone an ATM fan?' a
--
Jay R. Ashworth Baylink jra at baylink.com
Designer The Things I Think RFC 2100
Ashworth & Associates http://baylink.pitas.com '87 e24
St Petersburg FL USA http://photo.imageinc.us +1 727 647 1274
Those who cast the vote decide nothing.
Those who count the vote decide everything.
-- (Joseph Stalin)
|
Well in the real world, your hypothesis has been proven wrong.
Thanks,
Steve Totaro
|
To correct my previous statement. Who said anything about Asterisk
doing the recording function at all? Certainly not me. My recording
servers sometimes run Windows but usually Linux just to keep
everything in the "Phone Rack" on the same OS.
Thanks,
Steve Totaro |
|
| Back to top |
|
|
stotaro at totarotechn...
Guest
|
| Posted: Fri Apr 25, 2008 11:29 am Post subject: [asterisk-users] Quality problems with ISDN PRI |
|
|
On Fri, Apr 25, 2008 at 11:10 AM, Doug Lytle <support at drdos.info> wrote:
| Quote: | Steve Totaro wrote:
| Quote: | That is interesting. I have an intel C2D and I can only see two
procs, not four, is that normal? Are you sure what you are saying is
|
I believe Intel removed HyperThreading after it moved over to dual cores.
Doug
--
Ben Franklin quote:
"Those who would give up Essential Liberty to purchase a little Temporary Safety, deserve neither Liberty nor Safety."
|
My dual proc, dual core AMD boxen show as four procs. I guess the AMD
architecture uses Hypertheading (or whatever the equivalent is for
AMD, I assume Intel owns the rights to the name Hyperthreading).
Thanks,
Steve Totaro |
|
| Back to top |
|
|
support at drdos.info
Guest
|
| Posted: Fri Apr 25, 2008 11:55 am Post subject: [asterisk-users] Quality problems with ISDN PRI |
|
|
Steve Totaro wrote:
| Quote: | My dual proc, dual core AMD boxen show as four procs. I guess the AMD
architecture uses Hypertheading (or whatever the equivalent is for
AMD, I assume Intel owns the rights to the name Hyperthreading).
|
Not that I'm aware of.
But I did find this article from back in 2002:
http://www.geek.com/amd-to-do-hyperthreading/
Doug
--
Ben Franklin quote:
"Those who would give up Essential Liberty to purchase a little Temporary Safety, deserve neither Liberty nor Safety." |
|
| Back to top |
|
|
jsmith at digium.com
Guest
|
| Posted: Fri Apr 25, 2008 2:13 pm Post subject: [asterisk-users] Quality problems with ISDN PRI |
|
|
On Fri, 2008-04-25 at 18:48 +0000, Arthur wrote:
| Quote: | I still hope someone would enlighten us by his experience in doing
call recordings without recording to RAM Drive.
|
I can't speak for Steve's solution (as I'm not sure exactly what he's
doing) but I could take a stab in the dark and guess that he's capturing
the audio at the network layer (on a completely different box than
Asterisk is running on) and recording it from there. But that's just a
guess...
--
Jared Smith
Community Relations Manager
Digium, Inc. |
|
| Back to top |
|
|
astmattf at gmail.com
Guest
|
| Posted: Fri Apr 25, 2008 3:15 pm Post subject: [asterisk-users] Quality problems with ISDN PRI |
|
|
On 4/25/08, Jared Smith <jsmith at digium.com> wrote:
| Quote: | On Fri, 2008-04-25 at 18:48 +0000, Arthur wrote:
| Quote: | I still hope someone would enlighten us by his experience in doing
call recordings without recording to RAM Drive.
|
I can't speak for Steve's solution (as I'm not sure exactly what he's
doing) but I could take a stab in the dark and guess that he's capturing
the audio at the network layer (on a completely different box than
Asterisk is running on) and recording it from there. But that's just a
guess...
|
To address several points:
OrecX (http://www.orecx.com/) can do call recording outside of the
Asterisk core using several different methods depending on your needs
and channeltypes. In fact even with Sangoma TDM cards you can capture
audio at the kernel level and send the audio as RTP streams very
efficiently(3% CPU load for 92 channels) to an OrecX server on your
network. It must be mentioned that setting up Orecx with retrieval
might be a little complex for some Asterisk users, especially if you
are recording a large amount of calls, or are recording on more than
one Asterisk server, and if you choose this route you would do well to
hire an experienced consultant(or contact Oreca directly) to do the
install for you.
As far as Asterisk-based recording, writing to a RAM drive(or tmpfs)
is about your only option if you are planning on doing more than 50
concurrent recordings, if you are using Asterisk it is a viable and
tested solution. I have several client systems that are recording well
over 50 calls concurrently on a daily basis this way.
If you will be recording directly to hard drives with any frequency or
volume I would strongly recommend NOT using standard IDE or SATA hard
drives, they burn up and fast. Use a caching SCSI drive controller
with some high quality SCSI drives and you can record to those drives
for years even at 40 concurrent channels recording all day every day.
Hope that helps,
MATT--- |
|
| Back to top |
|
|
stotaro at totarotechn...
Guest
|
| Posted: Sat Apr 26, 2008 7:52 am Post subject: [asterisk-users] Quality problems with ISDN PRI |
|
|
On Fri, Apr 25, 2008 at 4:15 PM, Matt Florell <astmattf at gmail.com> wrote:
| Quote: | On 4/25/08, Jared Smith <jsmith at digium.com> wrote:
| Quote: | On Fri, 2008-04-25 at 18:48 +0000, Arthur wrote:
| Quote: | I still hope someone would enlighten us by his experience in doing
call recordings without recording to RAM Drive.
|
I can't speak for Steve's solution (as I'm not sure exactly what he's
doing) but I could take a stab in the dark and guess that he's capturing
the audio at the network layer (on a completely different box than
Asterisk is running on) and recording it from there. But that's just a
guess...
|
To address several points:
OrecX (http://www.orecx.com/) can do call recording outside of the
Asterisk core using several different methods depending on your needs
and channeltypes. In fact even with Sangoma TDM cards you can capture
audio at the kernel level and send the audio as RTP streams very
efficiently(3% CPU load for 92 channels) to an OrecX server on your
network. It must be mentioned that setting up Orecx with retrieval
might be a little complex for some Asterisk users, especially if you
are recording a large amount of calls, or are recording on more than
one Asterisk server, and if you choose this route you would do well to
hire an experienced consultant(or contact Oreca directly) to do the
install for you.
As far as Asterisk-based recording, writing to a RAM drive(or tmpfs)
is about your only option if you are planning on doing more than 50
concurrent recordings, if you are using Asterisk it is a viable and
tested solution. I have several client systems that are recording well
over 50 calls concurrently on a daily basis this way.
If you will be recording directly to hard drives with any frequency or
volume I would strongly recommend NOT using standard IDE or SATA hard
drives, they burn up and fast. Use a caching SCSI drive controller
with some high quality SCSI drives and you can record to those drives
for years even at 40 concurrent channels recording all day every day.
Hope that helps,
MATT---
|
I paid for the OrecX ability to save the recordings as the sipcallid.
This is fairly easy to track and match up in a CRM so long as you are
writing to a DB.
Before that you just had a bunch of folders based on day and hour and
the filenames were impossible to track, IP address and time I believe.
Not much use when recording 15k calls a day.
I also worked with Bruno @ Oreca to get their "passive recording"
solution from it's infancy (~10 or so concurrent calls) to a real
enterprise solution (maxing at ~200 concurrent recordings per server).
Thanks,
Steve Totaro |
|
| Back to top |
|
|
benny+usenet at amorse...
Guest
|
| Posted: Sat Apr 26, 2008 1:37 pm Post subject: [asterisk-users] Quality problems with ISDN PRI |
|
|
"Steve Totaro" <stotaro at totarotechnologies.com> writes:
| Quote: | My dual proc, dual core AMD boxen show as four procs. I guess the AMD
architecture uses Hypertheading (or whatever the equivalent is for
AMD, I assume Intel owns the rights to the name Hyperthreading).
|
I think the more likely explanation is that two times two is four.
/Benny |
|
| Back to top |
|
|
stotaro at totarotechn...
Guest
|
| Posted: Sat Apr 26, 2008 1:43 pm Post subject: [asterisk-users] Quality problems with ISDN PRI |
|
|
On Sat, Apr 26, 2008 at 2:37 PM, Benny Amorsen <benny+usenet at amorsen.dk> wrote:
| Quote: | "Steve Totaro" <stotaro at totarotechnologies.com> writes:
| Quote: | My dual proc, dual core AMD boxen show as four procs. I guess the AMD
architecture uses Hypertheading (or whatever the equivalent is for
AMD, I assume Intel owns the rights to the name Hyperthreading).
|
I think the more likely explanation is that two times two is four.
/Benny
|
But then that gets back to my Intel C2D show as two procs. 2 x 2 = 2.
Or is C2D not four cores?
Thanks,
Steve Totaro |
|
| Back to top |
|
|
gordon+asterisk at dro...
Guest
|
| Posted: Sat Apr 26, 2008 2:17 pm Post subject: [asterisk-users] Quality problems with ISDN PRI |
|
|
On Sat, 26 Apr 2008, Steve Totaro wrote:
| Quote: | On Sat, Apr 26, 2008 at 2:37 PM, Benny Amorsen <benny+usenet at amorsen.dk> wrote:
| Quote: | "Steve Totaro" <stotaro at totarotechnologies.com> writes:
| Quote: | My dual proc, dual core AMD boxen show as four procs. I guess the AMD
architecture uses Hypertheading (or whatever the equivalent is for
AMD, I assume Intel owns the rights to the name Hyperthreading).
|
I think the more likely explanation is that two times two is four.
/Benny
|
But then that gets back to my Intel C2D show as two procs. 2 x 2 = 2.
Or is C2D not four cores?
|
Have a look at the 'flag's in /proc/cpuinfo
If there's a 'ht' then it's a hyperthreading core.
Otherwise its a 'real' processor, and if there are N on a chip, then well
and good.
Hyperthreading is like having an extra quarter of a processor, depending
on what tasks you're doing.
But some BIOSes can disable the HT part of a core, and a modern kernel
ought to have hyperthreading support compiled in to make the best use of
it. (Another reason I always custom compile kernels for my applications)
As far as I'm aware, AMD hasn't made a "hyperthreaded" core, so it's
"real" cores on the same chip. So if the "dual proc, dual core" unit above
has 2 CPU chips, then it's 4 processors.
Going back to Intel: "Core 2" is Intels name for that particular family of
processors. The 2 in the name does not indicate the number of cores! So
you can have a Core 2 Solo - one core, Core 2 Duo - 2 cores and Core 2
Quad - 4 cores.
Gory details:
http://en.wikipedia.org/wiki/Hyperthread
http://en.wikipedia.org/wiki/Intel_Core_2
Gordon |
|
| Back to top |
|
|
rob at hillis.dyndns.org
Guest
|
| Posted: Sat Apr 26, 2008 5:10 pm Post subject: [asterisk-users] Quality problems with ISDN PRI |
|
|
No, a dual core processor has two cores. 
My Quad core shows four processors.
Steve Totaro wrote:
| Quote: | On Sat, Apr 26, 2008 at 2:37 PM, Benny Amorsen <benny+usenet at amorsen.dk> wrote:
| Quote: | "Steve Totaro" <stotaro at totarotechnologies.com> writes:
| Quote: | My dual proc, dual core AMD boxen show as four procs. I guess the AMD
architecture uses Hypertheading (or whatever the equivalent is for
AMD, I assume Intel owns the rights to the name Hyperthreading).
|
I think the more likely explanation is that two times two is four.
/Benny
|
But then that gets back to my Intel C2D show as two procs. 2 x 2 = 2.
Or is C2D not four cores?
Thanks,
Steve Totaro
_______________________________________________
-- Bandwidth and Colocation Provided by http://www.api-digital.com --
asterisk-users mailing list
To UNSUBSCRIBE or update options visit:
http://lists.digium.com/mailman/listinfo/asterisk-users
!DSPAM:48137bd6112601373216315!
|
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.digium.com/pipermail/asterisk-users/attachments/20080427/525083db/attachment.htm |
|
| Back to top |
|
|
benny+usenet at amorse...
Guest
|
| Posted: Sat Apr 26, 2008 5:39 pm Post subject: [asterisk-users] Quality problems with ISDN PRI |
|
|
"Steve Totaro" <stotaro at totarotechnologies.com> writes:
| Quote: | But then that gets back to my Intel C2D show as two procs. 2 x 2 = 2.
Or is C2D not four cores?
|
D is for duo.
/Benny |
|
| Back to top |
|
|
stotaro at totarotechn...
Guest
|
| Posted: Sat Apr 26, 2008 10:31 pm Post subject: [asterisk-users] Quality problems with ISDN PRI |
|
|
Got it, sorry, nothing to see here...
I thought the 2 and duo meant 2 x 2 as the name would imply (to me at least).
I stopped keeping up on processors with the exception of exceptional
hardware like the ARM, RISC, FPGA, and Itanium2.
Thanks,
Steve Totaro
On Sat, Apr 26, 2008 at 7:17 PM, Arthur <pylinuxian at gmail.com> wrote:
|
|
| Back to top |
|
|
andersen at mwdental.com
Guest
|
| Posted: Mon Apr 28, 2008 10:58 am Post subject: [asterisk-users] Quality problems with ISDN PRI |
|
|
| Quote: | But then that gets back to my Intel C2D show as two procs. 2 x 2 = 2.
Or is C2D not four cores?
|
If I'm not mistaken, there was a "Core Duo" - which was dual core processor.
Now there is a "Core 2 Duo" - which is a "second generation" dual core
processor.
Still just 2 cores though... (which would show as 2 processors in Linux, not
4)
Bill |
|
| Back to top |
|
|
|
|
|
You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum
|
Powered by phpBB © 2001, 2005 phpBB Group
|
Search
