On Fri, 15 Jul 2005, Yehuda Goldenberg wrote:

> Our company is looking to replace an antiquated phone system.

So far, so good.

> I was leaning towards using a traditional PBX, because I figured that it
> would be more reliable and it wouldn't be subject to the problems of IP
> networks.

That's a good direction to be leaning.

> One vendor is trying to get me to change my mind about that. He claims
> that we can keep the voice and data networks completely separate by
> running vlans. The IP phones have vlan switches on them and one wire can
> be run to each desk and the pc and the phone can be on separate vlans.

Let's hope the phones' "vlan switches" are 802.1Q compliant...

> I was concerned that problems on the data network such as viruses would
> still bring down the whole thing, and in addition to the pcs not
> working, the phones wouldn't work either and it would be total mayhem.
> He says that QOS would make sure that the phone calls always go through
> even if the data network is completely dead.

Then he has a very narrow definition of "completely dead". If a network
switch is Dos'd-to-99%-CPU or just plain crashed, then packets won't be
getting through regardless of the value of the QOS field.

> I was also concerned that the VOIP system would mean every call - even
> desk-to-desk would go through the internet

Probably true if the VOIP PBX is not on your premises.

> and if the T1 to the internet
> goes down, the phones don't work. His answer to that was redundant T1
> links,

Which probably will run over the same strand of fiber.

> and since they are the ISP and the VOIP provider, they will give
> us a reliable network that won't go down. Also all the VOIP equipment on
> their end is redundant.

You need more than redundant equipment to guarantee reliability. They
should also have redundant gateways to different POTS (Plain Old Telephone
Service) carriers (after all, 99% of the calls you'll be making will be to
phones on POTS).

I've been doing a little research on this subject because last week I
applied for a position at a university which is planning a
wholesale migration to VOIP (the position is with their TeleCom dept,
which apparently is recruiting in order to have some in-house IP expertise).

The position posting had some supplemental questions, one of which was
"What do you feel is the biggest challenge facing such a migration?"

I replied that their biggest challenge would be achieving comparable
reliability, and the biggest challenge to that is security. Traditional
telephony systems are pretty well isolated from those seeking to disrupt
them, and with VOIP you lose that inherent isolation. "It is a sad fact
that too many IT systems and networks are deployed without regard for
security considerations. Security is usually adressed after-the-fact, and
as a result is usually inadequate. Regardless of whether you select me for
your position, I hope you recognize the need to engineer-in security from
the start..."

Which pretty much guarantees I won't get the position, 'cause we all know
that most folks outside our profession regard ITsec as an inconvenient
obstacle to doing their job.

So I don't feel I'm risking much, by stating in this public forum:

Don't do it. Stick with a traditional (non-VOIP) PBX.



p.s. ran across this while looking into VOIP:


Scott L. Stursa 850/644-2591
Network Security Analyst stursa@mailer.fsu.edu
OTI Enterprise Security Group Florida State University

- No good deed goes unpunished -
firewall-wizards mailing list