[Asterisk-Users] ISPs with QoS for VoIP?

[Brian J. Schrock]

Actually I saw slides from a presentation by the CTO of Sprint two  
years ago discussing almost the exact opposite. The larger the pipe the  
less need or even opportunity to apply QoS. Ascii art.


T1

                              *       *    *  |   D
                     *                         |   e
                                                |   l
         *                                     |   a
  *                                            |   y
-----------------------------------
            Throughput
OC12

                                                   * |  D
                                                  *  |  e
                                                *    |  l
                                   *                 |  a
*                                                   |  y
_______________________
              Throughput

Rules learned with narrow band networks and chatty LAN protocols lose  
relevance as b-width increases not because of total throughput, but  
because of the frequency the link operates at. QoS is nothing more than  
a ploy by those with narrow band business models to create artificial  
scarcities in an effort to create new sources of revenue. QoS was being  
pushed really hard when ATM was hot, and who were the people really  
excited about ATM (At least what I saw)? All those dinosaurs with  
narrow band business models. Do not allow QoS to be built into the  
network, b-width should be a commodity.

On Saturday, March 8, 2003, at 11:54 AM, Roderick Montgomery wrote:

> According to alex at pilosoft.com:
>> The only case where QoS is useful is on tail-end circuits. Everywhere
>> else, having bigger pipes is much more preferable to QoS.
>
> Um, not to state the obvious, but aren't bigger pipes ALWAYS  
> preferable?
>
> QoS does not make packets move faster.
>
> Please reread the previous sentence.
>
> QoS implementations essentially tell network elements (routers,  
> switches,
> etc.) which packets may be slowed or dropped entirely relative to other
> packets. I think everyone here can agree that slow or failed delivery  
> of
> packets is a Bad Thing. However, since the "bigger pipes" solution  
> usually
> carries a monetary cost, it's not always available.
>
> QoS is a "damage control" technology. If your network needs outstrip  
> your
> network capacity, QoS is one way to deal with the problem: attempt to
> de-prioritize some traffic to allow more capacity for your voice  
> traffic.
>
> If a network segment or element becomes congested, there are many other
> options available -- remove the competing traffic altogether. Complain  
> to or
> change your ISP (if it's their problem). Segment your voice traffic  
> with
> VLANs.
>
> If your network is already congested, QoS is a way to direct the  
> effects of
> congestion at the traffic that is less important to you. It is not a  
> way to
> make "good" traffic go faster.
>
> rm
> ----------------------------------------------------------------------- 
> --
>  Roderick Montgomery   rod at thecomplex.com    
> <URL:http://thecomplex.com/>
> the fool stands only to fall, but the wise trip on grace... [Sarah  
> Masen]
> ----------------------------------------------------------------------- 
> --
> _______________________________________________
> Asterisk-Users mailing list
> Asterisk-Users at lists.digium.com
> http://lists.digium.com/mailman/listinfo/asterisk-users
>
>
Brian J. Schrock
Network Engineer, RHCE, CCNA
Anistone Technologies
Phone: 614-537-2817
FAX: 614-573-7165
6926 Avery Rd.
Dublin, OH 43017