There is no term in telephony, and perhaps the English language, that offers more hype, interpretations, or confusion than "the cloud". It is clear the cloud is going to change things, with new models, new prices, and new capabilities. Many of the definitions, practices, and concepts of cloud computing do not apply to cloud telephony.
For those considering cloud telephony, a reasonable question is which of the multitude of communication options today are indeed cloud solutions. It isn’t a silly question because just about every product, service, and solution claim to be cloud. In fact, based on current marketing, it’s pretty hard to avoid the cloud. This post is not titled "Why Cloud Telephony?"--that's an entirely different conversation.
Defining the cloud is a stumbling block. Many feel it's a pricing matter, some insist on specific technologies, others assume APIs. Let me start by proposing what isn’t cloud telephony: basic carrier services including land line dial tone and wireless carriers. Carriers are not new (they just merge a lot), the technologies change but telecom's simple base requirement for a network does not make all telephony cloud.
Another non-cloud solution is co-location. This involves a PBX (or similar server) residing in a remote data center. The cost and management of the equipment remains a private/internal matter. Co-location is an important notion as it represents a partial step to the cloud. Organizations that embrace co-location realize the equipment no longer needs to be on-premise. To users, there may be little perceived difference between cloud services and co-location. However, co-location is really more of a facilities matter.
The National Institute of Standards and Technology (NIST) defined cloud computing, but telephony doesn't quite fit it. Most definitions of cloud telephony include two critical elements: a service, not a product; and a utility pricing model. Both of these are contentious. A service can be the actual application (telephony) or the infrastructure. Utility pricing means it is based on actual usage (such as electricity). Some services are metered, some are offered as a flat monthly price--that isn't important.
What is important is the cloud tends to have negligible customer start-up costs or long term commitments; even that is subject to debate. In CFOspeak, cloud services tend to be Opex instead of Capex. Other popular descriptors include web based, device independence , rich APIs, scalable, and many more.
What follows are the conceptual models submitted as my exhaustive list of how to cloud telephony.
1) Hosted Voice
Dial tone (POTS or IP) is not the cloud because it’s basic telephony. However, hosted voice crosses over as it represents enhanced services. Frequently described as a virtual PBX, hosted voice crosses the blurry line into the cloud. Providers include companies like Simple Signal, Packet8, and good ol' Centrex from the RBOCs--they tend to use labels like SaaS, Hosted VoIP, and Hosted UC. The basic concept is a virtual telephony solution offering advanced features typically associated with a PBX, delivered as a service.
2) Cloud Applications
Using the cloud to solve for specific applications is an approach rapidly growing in appeal. An answering service or Directory Assistance are the classic outsourced telephony apps, but now it is possible to use automated cloud services for communications enabled business processes (CEBP), and doing so does not necessarily require a new phone system. Companies such as IfbyPhone, Voxeo, and Twilio offer comprehensive tools for custom or off-the-shelf communications solutions.
With this model, an organization can voice- (or SMS-) enable sophisticated and specialized applications to update databases, initiate calls, manage contest voting, update or query databases, and much more without the need to purchase specialized equipment or capacity.
"Cloudbursting" to solve for capacity spikes also falls into this category. Traditional telephony involves capacity planning around the busiest "hour" of the busiest day of the busiest week of the year. Since many cloud services use utility pricing models, it can be cost-effective to use cloud services to manage peak demand levels or even as part of a disaster recovery plan.
3) Virtual Numbers
Virtual numbers are a simple concept, but can be tricky as it’s a voice service that requires users to obtain separate voice service. Examples of virtual number services include Ringio, OpenVBX, and Google Voice. These services provide phone numbers that route (or not) to other numbers based on user defined criteria.
Virtual numbers offer a variety of enhanced services, such as the ability to ring multiple phones, and various tools of unified communications. Virtual numbers associate the features with the numbers rather than the phones or phone systems--and promote portability and mobility. The category is fairly young, but expect significant business class innovation in this area.
Unlike most cloud models, virtual number services don't rely heavily on the Internet. In fact, virtual numbers can be combined with (or replace) traditional telephone systems and/or cell phones to create a reasonably feature-rich UC solution.
4) Private Cloud
Private clouds are where ‘the cloud’ gets hazy and offensive to many cloud purists. The cloud enabling technologies are developing and gaining the interest of many CIOs--particularly in connection with virtualization. Virtualizing internal infrastructure creates a computing ‘service’ that can be adjusted and managed easier than separate dedicated servers. Organizations that deploy private clouds leverage cloud technologies to reduce costs and increase availability and retain full control.
Organizations can realize many benefits and efficiencies with private clouds--albeit not utility pricing. Avaya, Siemens, Mitel, and NEC offer voice solutions that run in virtualized environments. This is fairly cutting edge as real time applications such as telephony were not compatible with virtualization technologies until recently.
5) Public Cloud
There is considerable discussion on the merits of private clouds versus public clouds within cloud computing camps. But voice is not yet a mainstream solution for public clouds. That will likely change. Certainly voice servers are being implemented in the public cloud, but rarely at the enterprise level. The big issue was the nature of real time processing, but as mentioned progress is being made there.
The second issue is most cloud providers require the use of the public Internet for access, raising QoS issues. Cloud providers are enhancing available networking options. Recently, Amazon announced customer improved control over IP addressing, and Hosting.com now supports MPLS and other private networks to its cloud offering. Siemens Enterprise continues to test a public cloud offering of OpenScape in Amazon’s cloud.
6) Cloud Appliance Hybrid
This approach is an evolution of managed services where a blackbox appliance is implemented on-site as a fully managed extension of a cloud platform. The concept has been around (Google offers a search appliance with this architecture), and there are indications it will expand to telephony. Microsoft recently announced an Azure appliance for large customers. It won’t be voice-ready on its first version, but it’s likely on the road map. Mitel’s upcoming AnyWare solution offers on/off site computing. IBM's Foundations product targets SMB users and comes with a variety of cloud services including management and disaster recovery services.
Conclusion
The cloud isn't for everyone. On the other hand, the PBX (though no one calls it that anymore) is tried and true, but it's no longer for everyone either. Slightly confusing these models is the shift toward intelligent endpoints at the network edge. Traditional telecom networks used simple endpoints and big honking switches for routing and services. Skype, and other services, are pushing the intelligence from the network out to the edge, enabling peer to peer services. This shift is fascinating, and further complicates cloud boundaries. Paradoxically, the cloud to some extent represents a strategy of distribution or decentralization. This is demonstrated practically with mash-ups such as using Google Maps or social network APIs with callerID.
The Internet has introduced a level of connectedness that enables new types of information and services. This will continue to impact telephony with many new and exciting capabilities and expectations, as innovation continues to ramp up.
Dave Michels, principal of Verge1, is a regularly contributor and blogs about Telecom at Pin Drop Soup.
