[SIPForum-techwg] TCP vs. UDP (SP adoption of TCP?)

Hiers, David David_Hiers at adp.com
Tue May 20 13:27:34 EDT 2008


All,
As a service provider using an expensive mix of best-of-breed SIP
implementations that service a fair number of endpoints and minutes, I
see the ability to use UDP as one of SIP's primary weaknesses.  

The attempt to supplant the TDM PSTN five-nines reliability is not
particularly well facilitated by running the entire show over an
unreliable transport protocol whose primary benefit is reduced HW
resource requirements.

When shopping for future peering partners, I would tend to prefer one
who's engineering philosophy happily embraces the deterministic control
of TCP rather than the fire-and-forget nature of UDP.

Best regards,


David Hiers

CCIE (R/S, V), CISSP
ADP Dealer Services
2525 SW 1st Ave.
Suite 300W
Portland, OR 97201
o: 503-205-4467
f: 503-402-3277 



-----Original Message-----
From: techwg-bounces at sipforum.org [mailto:techwg-bounces at sipforum.org]
On Behalf Of Russell Bennett
Sent: Tuesday, May 20, 2008 9:21 AM
To: Zweig, Greg; techwg at sipforum.org
Subject: Re: [SIPForum-techwg] TCP vs. UDP (SP adoption of TCP?)

Greg et al.

I have already agreed that we have to accommodate the past, but we also
have to drive the future.  If we "make it OK to just offer UDP" then
that is what people will think that it is acceptable to do.

I have been actively engaged with a number of service providers for
quite some time now and, without naming names or providing timelines, I
can assure you that TCP is being deployed by service providers.

As for encouraging quick adoption - the issue is not one of which SIP
Transport is best/cheapest/already there; the issue is one of simple
economics: SIP Trunks cannibalize Service Provider revenues under
existing pricing models.  The SPs who offer "UDP only" are seeing little
demand for SIP Trunking - what they don't realize is that it is a
"chicken and egg" problem.

We must move SP SIP connectivity capability forward such that SPs can
start to deploy revenue generating services and applications to replace
revenue lost by decrementing PRI trunks.  We can only get to a service
federation model by moving rapidly beyond SIP Voice.  If we can't *even*
get to more broadly deployed SIP Voice, then (some) SPs will remain
where they are - stuck in the 20th century.

Forward looking SPs that I work with have heard this loud and clear.

UDP should be a 'MUST' to support older architectures.  TCP MUST be a
'MUST' to support newer architectures.  I don't see how we accommodate
the past and drive the future any other way.


Russell

-----Original Message-----
From: techwg-bounces at sipforum.org [mailto:techwg-bounces at sipforum.org]
On Behalf Of Zweig, Greg
Sent: Monday, May 19, 2008 11:39 AM
To: Peter Dunkley; techwg at sipforum.org
Subject: Re: [SIPForum-techwg] TCP vs. UDP (SP adoption of TCP?)

Agreed -- We also have to consider the capabilities at the customer
premise.  I wouldn't want to advocate a solution that drives end
customers into a position where their existing equipment is made
obsolete.  There are many products in-use that could be upgraded to
other facets of a SIPconnect 1.1 implementation but will never cost
effectively support TCP.  I want to embrace them, not disenfranchise
them.

We need to accommodate the past while we drive the future.

Greg Zweig
Sonus Networks. Inc
(978) 614-8027


-----Original Message-----
From: techwg-bounces at sipforum.org [mailto:techwg-bounces at sipforum.org]
On Behalf Of Peter Dunkley
Sent: Monday, May 19, 2008 2:23 PM
To: techwg at sipforum.org
Subject: Re: [SIPForum-techwg] TCP vs. UDP (SP adoption of TCP?)

The danger with this approach is that you will may end up with a
wonderful SIPconnect 1.1 recommendation that the writers/purists love
but is unpalatable to the industry.

There needs to be a pragmatic approach taken that illustrates what SIP
trunking should be, while still encouraging quick adoption.

Making SPs have to choose between changing their architecture or
becomming compliant will not encourage them to adopt SIPconnect (quickly

or otherwise).

Russell Bennett wrote:
> All,
>
> I know from trying to work with the SPs on SIP Trunking that *some*
SPs are going to take a firm position on the support of UDP only.   The
question that Janne poses is:
>
>         1) Are we going to define SC1.1 to be what a SIP Trunk service
should be (e.g. UDP *and* TCP, among many other things)?
>         OR
>         2) Write SC1.1 to be what we think will enroll more SPs in the
SC compliance program?
>
> I argue strongly against 2 - this would not be a step forward on
SC1.0.  3261 is very clear on TCP support.  If SIP Forum was a
commercial entity, then we would go for 2 to maximize revenue.
>
> However, SF is a standards body and it is incumbent upon us to develop
SC1.1 as a *correct*, albeit pragmatic, document.  The key is the
balance between pragmatism and correctness and I, for one, think that
accommodating SPs that are mired in the 20th century is the wrong side
of that balance.
>
> Russell
>
>

_______________________________________________
techwg mailing list
Send mail to: techwg at sipforum.org
Unsubscribe or edit options at:
http://sipforum.org/mailman/listinfo/techwg

_______________________________________________
techwg mailing list
Send mail to: techwg at sipforum.org
Unsubscribe or edit options at:
http://sipforum.org/mailman/listinfo/techwg

_______________________________________________
techwg mailing list
Send mail to: techwg at sipforum.org
Unsubscribe or edit options at:
http://sipforum.org/mailman/listinfo/techwg


This message and any attachments are intended only for the use of the addressee and may contain information that is privileged and confidential. If the reader of the message is not the intended recipient or an authorized representative of the intended recipient, you are hereby notified that any dissemination of this communication is strictly prohibited. If you have received this communication in error, please notify us immediately by e-mail and delete the message and any attachments from your system.



More information about the techwg mailing list