[ILUG] Dilemma with open-source VOIP systems
chris.higgins at darach.ie
Wed Apr 7 14:05:13 IST 2004
On Wed, 07 Apr 2004 12:33:26 +0000
Denis Hennessy <denis at hennessynet.com> wrote:
> They're also proposing all kinds of infrastructure upgrades
> (MPLS-capable switches and routers to improve QoS) which I'm not
> completely convinced I need.
All that QoS will do for you is allow you to discriminate against
traffic. Identify the traffic that is not important, and put it
to the back of the queue. ( it's not dressed up like that, but
that's what it ends up doing ).
> I'd really like to hear from anyone that has tried an open-source
> * How did it work? Users happy?
> * Who did you get to install/support it?
> * Did you need to upgrade your IP infrastructure?
Does your IP infrastructure have sufficient *spare peak* bandwidth to
carry your predicted additional voice requirements ?
If you do have loads of spare bandwidth at peak times, then adding
QoS will probably slow things down ( the newer / faster devices
will have the least impact, but older slower devices may have a
significant impact). The slowdown may be noticable, then again
it may not - it's nigh on impossible to make these kind of
generalisations in any way specific.
When consultancy companies spec solutions for technologies
they do not fully understand ( either because they don't
know their stuff, or more commonly because they can't
see the business justification in understanding your
exact needs ) then the solution will be over spec'd
to make sure that it meets the needs of the customer
and allows headroom for growth.
If you know your traffic patterns, and can predict
your growth/usage fairly accurately then you should
be able to make some sort of accurate calculation on
how much traffic you need at any one time and then
the decision to upgrade/not upgrade becomes easy.
Everyone's network is different ( everyone has
different priorities on traffic )
Are your regional sites connected via private circuits
or the internet - 'cause if it's the internet then
my first ( wild ) gut reaction is that mpls/qos
in each regional site is probably a waste. You have got to
look at QoS end-to-end or not at all. What's the point
in having all the equipment inside your site working
overtime to discriminate against http traffic so
that the voice traffic can go first - if you then pass
that traffic to your isp and they go do their own
thing with it ( probably fifo or wred )
> Irish Linux Users' Group
Chris Higgins Cisco Learning Partner
Darach Technology Ltd tel: +353-1-6204370
email: chris.higgins at darach.ie fax: +353-1-6204371
More information about the ILUG