Pros and Cons for VoIP
By William Flanagan
Lest we forget, most of us in networking operate on the basis of certain
assumptions. We don't always articulate them to newcomers, or even to
ourselves very often. Some key assumptions about VoIP deserve a fresh
look regularly. So here's an airing of the arguments for and against
Voice over IP in the enterprise, many of which are taken for granted by
The original claim for VoIP, when it was strictly between PCs, was "free
phone calls." That's still true to some extent, but with long distance
now included at a flat rate, and seldom more than $.05/minute, "free"
doesn't mean much any more.
The big value in VoIP now appears in the form of new features that are
either harder, more expensive, or impossible to get from older TDM phone
switches. For example, conference calling, increasingly popular as
travel becomes less attractive, can be quite expensive when outsourced.
Most VoIP softswitches include conferencing for near zero incremental
cost--assuming the network infrastructure can support this service. An
enterprise can save hundreds of thousands of dollars per year on this
Most of the new features fall under a convenient header, Unified
Communications (UC), including:
-- receiving faxes and voice mails as attachments to emails (making them
easier to sort through and discard),
-- getting information services (directories, customer stats or pictures
of the caller displayed on the phone, etc.),
-- having a single number for all your phones (calls forking to all
-- finding enhanced mobility from the hand-off of calls between Wi-Fi
and cellular networks (reducing cellular airtime minutes),
-- sharing Instant Messages, texting, whiteboards, document, and
other services while talking on the phone,
-- letting customers "click to talk" directly over the Internet from
At a recent Siemens seminar, UC was predicted to be practically
everywhere in 3 to 5 years. The results will probably surprise us with
new ways to collaborate with colleagues, customers, and vendors.
Once the legacy TDM equipment is gone, there will be little or no need
for TDM circuits and trunks. Consolidating all carrier access on large
IP transmission lines will reduce costs there. And when voice is
another form of IP traffic, a desktop will need only one LAN cable--or
perhaps none if the site is wireless. Moves, Adds, and Changes of
telephones can be automated, saving big bucks in locations where people
move around. However, we're unlikely to reach this goal for many years.
A unique feature of VoIP is "presence," the ability of a system to keep
users informed of the availability and status of all other users.
Should save time by avoiding wasted jaunts down the hall to an empty
office, but most of all by finding someone quickly.
Presence. As in "Good news, bad news." Don't you know people who hide
behind voicemail? How can they do that when everyone can see if they
are logged in, on the phone, etc. There will be resistance, which may
extend to the whole concept of VoIP.
Cost comparisons vary enormously, from VoIP costing more than a PBX to
saving over $12 million (Siemen's internal example). As we consultants
say, "It depends." Some VoIP vendors can support 100,000 phones on two
servers that back up each other. Other systems would require dozens of
servers for the same capacity and redundancy.
You can expect relatively high costs for fancy desktop phones, mobile
handsets that combine Wi-Fi and cellular (though these will get cheaper
soon), and possibly for upgrades to the network to provide
prioritization (QoS) for voice traffic and Power over Ethernet to the
fixed phones. PoE allows backup power in the wiring closets to keep
all the phones running for some time after a power failure. Reputable
VoIP vendors won't sell the servers and desk sets without confirming
that the network infrastructure can deliver good voice quality--and that
may cost you.
When voice is on the data network, voice availability may be affected
adversely by common practices of LAN technicians, including rebooting
servers to see if that clears a problem, taking servers down to patch
software, and powering down to repair hardware failures. You can avoid
interruptions to voice service through redundancy, proper configurations,
and appropriate procedures, all of which can be expensive in hardware,
software, and training.
Voice signaling protocols (SIP, H.323) and the streaming protocols for
content (RTP, SRTP) were not designed for Network Address Translation
(NAT). If VoIP connections pass through a NAT function, there must be
additional services offered outside the firewall perimeter to allow
direct-dialed incoming calls to reach the proper phone. Servers or
specialized appliances support this function, for example Session Border
Controllers. But that means that some voice servers are connected to
the Internet, potentially exposed to attack. After all, call processor software
is an application running on a server--with all the vulnerabilities of the
hardware, operating system, as well as those applications.
VoIP security is not under complete control yet; it may never be
better than security for other server farms. Until we know more and
have greater confidence in securing voice services, some enterprises are
avoiding the Internet connection by transitioning all off-site VoIP traffic to
legacy PSTN formats in a gateway (VoIP on one side, ISDN or POTS trunks
on the other side). This approach definitely blocks hackers, but also
cuts off many of the features in UC.
At least with POTS or ISDN trunks at each site the requirement to
provide information with 9-1-1 calls is easier to meet. When VoIP
needs to meet enhanced 911 (E911) specs, the function that tracks phone
locations will need to be very sophisticated so it can track to the
floor/quadrant or even to the cubicle/office. Fortunately, third-party
services can help here, but at a cost.
Have an opinion? Let us know at firstname.lastname@example.org.
| In Converged
We Have the Experience