[Asterisk-biz] DS3000P - 16 E1 capacity on single card (Are we there yet or temp solution)

[trixter aka Bret McDanel]

On Wed, 2005-11-23 at 21:27 -0500, Andrew Kohlsmith wrote:
> On Wednesday 23 November 2005 12:27, alex at pilosoft.com wrote:
> > b) savings from IAX trunking are overstated. On serial links, enable RTP
> > header compression for best savings...
> 
> Care to explain how the trunk savings are overstated?  Or better yet, have 
> some test data that shows the bandwidth of say a dozen SIP calls between two 
> systems with RTP compression, and a dozen trunked IAX2 calls between the same 
> servers?

This is a sticky subject becuase of many variables.  Look at ATM
framing, it uses a 53 byte fixed cell size, 5 bytes are header and on
the last packet in the cell an 8 byte SAR trailer is added.  If you use
pppoe you have an additional 6 bytes per packet added.  Becuase its
fixed size padding is added if you dont fully fill a cell.  packets are
split into those cells, so codec selection and sample size can affect
this ...

With IP+UDP+RTP you normally have about 40 bytes used.  RTP compression
can drop that 2 bytes (the IP, UDP and RTP headers are compressed
together).  Of course for this to work you have to have the remote side
of the link support it.  This leaves 41 bytes for payload and SAR
trailer for the first cell.  If you jam a G.726 packet in there at 9.5ms
sampling it will take exactly 1 ATM cell per sample sent.  Better than
padding it and effectively wasting that bandwidth.

IAX trunking allows multiple packets to the *same* destination be
combined with only one IP+UDP header.  This does provide some savings
however most calls are not to/from the same provider in residential
service, they can be with business service where many calls are placed
at the same time, and some savings can be realized there.  Overall less
bandwidth to traffic (the headers can be as much as 50% of the total
traffic, take G.726 at 10ms samples, 40 bytes payload 40 bytes headers,
add in 8 bytes of padding to fill the cell ...)

I think in some circumstances IAX trunking can yield some savings in
bandwidth (28 bytes per packet sent), however if you are fielding the
vast majority of your calls from different end points you wont see any
savings.  Perhaps that is what the original author meant by overstated
(although I would be interested in hearing his reasoning if it differs).

If you can also do CRTP (ip,udp,rtp are all compressed from 40 down to 2
bytes) on top of all of that then you would see far more savings.  


-- 
Trixter http://www.0xdecafbad.com     Bret McDanel
UK +44 870 340 4605   Germany +49 801 777 555 3402
US +1 360 207 0479 or +1 516 687 5200
FreeWorldDialup: 635378
http://www.sacaug.org/ Sacramento Asterisk Users Group
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: This is a digitally signed message part
Url : http://lists.digium.com/pipermail/asterisk-biz/attachments/20051123/6e363666/attachment.pgp