|
Development, By Definition
BY KEVIN MAYER AND CHRIS DONNER
[Go right to Service Creation
Made Easy Through Open Access]
At the risk of sounding like the hosts of a late-night paid programming
show or some slick TV evangelist, we decided to start off this article
with a definition. Or rather, with a question that requires a definition
for an answer. What exactly is service creation, and further, what changes
might we see in service creation as a result of the decomposed next-gen
network model? In pursuit of a definition, and more importantly an
understanding, we spoke with a number of vendors in the softswitch,
application server, and service creation space. Let's begin.
Robert Newcomb, VP corporate marketing of Ennovate Networks, perhaps
put it most directly and succinctly when he said, "Service creation
involves the establishment of a policy and then the enforcement of that
policy." This definition is effective, not only because it is short
and pithy, but because it points out the inherent benefits of the next-gen
network architecture: it doesn't mention transport.
STANDING ON TWO FEET
The key feature of the next-gen network, and the one that the softswitch
architecture seeks to capitalize upon is that transport and call
processing are separate and can be altered or upgraded independently of
one another. Think of two of an automobile's most basic systems, the
engine and the suspension, and then think of the automobile's most basic
function. Your steering and your use of the accelerator are coordinated to
deliver you from point A to point B, but the two systems are completely
discrete and self-contained. If you go to a mechanic with a steering
problem -- the car pulls hard to the right, for instance -- there is no
need to look at the engine or power train. Instead, the mechanic will
focus on the suspension.
Yet, no matter how discrete these systems are, they still operate very
much in tandem. No matter how much you steer, you will never arrive at
your destination if you are out of fuel. With the next-gen network this
factor is a significant point of differentiation among the many equipment
vendors working in this space. As Carl Baptiste, managing director of
product management, Convergent Networks said, "In the midst of this
move to decouple services from transport, there is a certain level of
interdependence between the layers that must be recognized as well."
In other words, service creation must always affect both transport and
processing.
The question of the moment, and the place where next-gen equipment
vendors are striving to differentiate themselves, is: To what extent must
both processing and transport be involved in service creation? There are
essentially two camps on this issue, and their products, white papers, and
industry positioning reflect the differences between them.
APPLICATIONS-ENABLED
Today, when you place a phone call, you do two things. First, you send out
a signal identifying yourself and the party you are calling to the network
and tell it to establish a connection. Second, you send your voice out
over that network and expect to hear someone else's voice (hopefully your
intended party's) coming back at you from out there somewhere. Cell phone,
home phone, office phone -- this is the basic function that the PSTN
performs for you, and it performs it flawlessly 99.999 percent of the
time.
Now, suppose you want the network to do more for you. Say, save
messages from callers who didn't manage to reach you because you were
online, or try you at your cell phone if you haven't answered your home
phone in three rings, or send you an e-mail at work telling you that your
good friend just called to cancel dinner tonight and wondered if you'd
like to meet for a drink instead and then allow you the option to either
e-mail or call back. Well, you have two choices. You could pick up the
phone and dial some obscure code like *475#36*33#*7 (funny... that looks a
lot like the swearing you'll be doing as you try to remember these
numbers). Or you could move up to the next-gen network.
Softswitch manufacturers are betting that enough people will want such
enhanced services that it will be in the best interest of service
providers (SPs) to roll them out quickly. The applications-enabled
softswitch manufacturers are betting on the service providers wanting to
provide some key enhanced services very quickly for certain markets,
quickly enough that they will be interested in a tightly integrated,
ready-to-deploy platform that presents both processing functions and an
applications server.
Applications-enabled softswitches also have an advantage in that they
can recognize and locate users at all times, without requiring
intelligence on the user's device. John Weald, VP of software engineering,
and Laura Thompson, VP of marketing and business development, Sylantro,
perhaps said it best: "True applications-enabled softswitches can
manage the overall call flow and features that can be invoked at any point
in the call flow. This allows many new possibilities to emerge."
These new possibilities can be bundled together by SPs and presented to
end users as vertical packages for a particular set of needs: a hotel
package, a call center package, an international enterprise package, and
so on.
Sounds good so far, right? So what are the drawbacks? Well, the tight
integration of the applications server and the softswitch, as much as it
results in greater control and flexibility, also tends to limit the
possibilities for outside control of the features. As John and Laura of
Sylantro put it, "We don't anticipate extending our development tools
to outside developers in the near future because of the 'mission-critical'
nature of our product."
And consider what Bill Leslie, CTO of LongBoard, had to say about
developing on such a switch: "Application-enabled softswitches may
employ proprietary links between the application layer and the call
control (softswitch) layer. While they may offer an API for interfacing to
the softswitch layer, the number of application providers willing to
program to that interface will
be reduced."
This is not to suggest that new services cannot be created -- not at
all. In fact, the Sylantro Communications Suite employs XML and Java Beans
to allow for development of new features. It's just that the linking of
the two functions naturally creates an environment that is somewhat less
open to third parties, in turn linking the SP to the vendor a bit more
permanently.
APPLICATION PLATFORMS
Almost diametrically opposed to the approach taken by the
applications-enabled camp is that taken by the manufacturers of
application platforms. Are they really that different? Well, consider this
from Jonathan Rosenberg, chief scientist with dynamicsoft:
"Softswitches are high-capacity, scalable devices, providing VoIP
connectivity to SS7 networks. Coupling access and services is generally
not a good idea." Doesn't sound like something you're likely to hear
from Sylantro, now does it?
In contrast to the applications-enabled approach, the application
platform approach insists that application servers should be completely
separate from the softswitch, and that they should be made available to a
wide range of developers through tools such as SIP, CGI, and XML.
Dan Dearing, VP of marketing, and Sridhar Ramachandran, CTO, of NexTone
Communications, discussed how this model allows independent software
vendors (ISVs) to play a role: "Using open interfaces, ISVs can focus
on the end-user application and be insulated from the details of managing
network connections. APIs (such as JAIN and Parlay) will be used to
establish network-based policies, and scripting languages (e.g., CPL, XML,
SIP-CGI) will likely be used by ISVs to create applications for network
clients such as PDAs and other IP devices."
Wow... sounds great, doesn't it? Now, for the drawbacks. As Carl
Baptiste of Convergent (an applications-enabled player) points out,
"The architecture required to support 'pure play' service creation
environments causes significantly more delay by requiring additional
messaging to take place between the softswitch and the application server
to activate the service."
Also, a model that suggests complete separation between the softswitch
and the application server requires more intelligence to reside in the end
user device to perform some enhanced services beyond basic Class 5
functions. While more and more SIP phones (and other intelligent end
devices) are available, some vendors and SPs are hesitant, suggesting that
it is unlikely that the millions of analog phones currently deployed are
going to be completely replaced with more intelligent devices anytime
soon.
This is certainly something to think about... although perhaps the real
reason for concern should be more environmental than anything. After all,
users have shown a tremendous willingness to upgrade cell phones yearly or
even more frequently, and often these phones are given out free with a
long-term service contract. The question is, if all the analog phones are
going to be replaced, where will they go to die? One more example of
fuller landfills through technology.
CREATION, OR CUSTOMIZATION
A final question that we posed to both camps of softswitch vendors was
regarding the types of new services to be created. After all, everyone is
talking of the killer app that will drive the next-gen network. What will
that app be like? Where will it come from?
The answers, across the board, were less than glamorous. A few
interviewees went so far as to deny that there would be any killer apps
created at all. Instead, they saw the killer being the combination of
existing services in new and exciting ways, like the earlier example where
you can reply to a voice mail using e-mail or choose to receive the voice
mail as e-mail. That's certainly a nice little feature, but is it killer
enough to convince someone to shell out more money per month, or to stay
with a service provider who is not offering the rock-bottom price? Perhaps
only time will tell.
Others expected a bit more on the development side, although almost
everyone we spoke to used the terms customization and creation
synonymously. Again, Robert Newcomb of Ennovate put it most directly:
"In the majority of cases, service creation will primarily be a
matter of customization and configuration..." Certainly easy-to-use,
open interfaces and platforms allow for lower cost configuration, self
provisioning, Web integration, and so on, and this is nothing to dismiss
out of hand. But for now it appears that the next-gen killer app might
just be a couple of last-gen apps cleverly combined.
A DEVELOPER'S CAVEAT
Rhetoric surrounding the open, next-gen network speaks of opening up
development to the wealth of Web and PC developers, taking advantage of
the brain power and creativity of a much larger development community than
traditional telecom can provide. Is it only rhetoric? After all, if
customizing and configuring existing services qualify as service creation,
then an in-house network manager is a developer. Sure, someone will have
to write the code to present the new GUI to the end user, but is that
really development? Where is the place for third parties?
The companies we canvassed had a wide range of answers to this
essential question -- so broad as to suggest that no one is really sure
what the opportunities will be here for third parties, or even if there
will be opportunities.
If no one is really sure, then why the rhetoric that claims a vast new
land of opportunity, just waiting for the homesteading Web developer?
Well, partly because no one is sure, and it's perhaps human nature to be
optimistic instead of pessimistic when confronted with uncertainty of this
magnitude. But also partly because no one can say any differently right
now, and it certainly sounds good to dangle this carrot in front of
regulatory agencies and consumers frustrated with paying $20 a month to
their ISP, $45 a month to their cable company, $60 a month for phone
service and long distance, and $60 a month for cell phone service. Or
businesses paying for both datacom and telecom services and unable to do
things like link cell phones, office phones, and e-mail addresses.
The promise of the next-gen network is that one day a miracle of
services will be possible through any one of the several service providers
that most people pay regularly today. In some ways, the disagreements
among softswitch manufacturers mimics the old differences between the
telecom and datacom networks. Applications-enabled softswitches are more
like the telecom network, with its insistence upon reliability and a more
controlled development environment, whereas the applications platform
softswitches are more like datacom, with its instability but exponential
growth and possibility.
The thing to watch as data and voice converge is whether SPs will be
able to maintain their five 9's reliability and still take advantage of
truly open and accessible network equipment and development. This ability
will decide the true definition of service creation in the next-gen
network.
Kevin and Chris would like to thank the following vendors and their
representatives for their help. While we may not have agreed with
everything they said, and while we certainly elaborated on what they
provided, without all of their input this article would have been much
less thorough: Ellacoya Networks, Sylantro Systems, NexTone
Communications, Ennovate Networks, Convergent Networks, Westwave
Communications, LongBoard, dynamicsoft, Ubiquity Software Corporation.
[ Return
To The March 2001 Table Of Contents ]
|