Cloud infrastructure for VoIP—are we there yet?
By Micah Singer, CEO, VoIP Logic
Every telecommunications carrier and service provider that is facilities based seems to have a huge amount of servers, networking devices, power distribution units and other equipment that takes up considerable space and uses considerable power. This gear needs to be networked, maintained and replaced on regular intervals and, in sum, is a capital and operational drain.
With the introduction (and promise) of production cloud systems—starting most notably with Amazon’s Elastic Computing cloud (EC2) in 2008—there has been a concerted effort to shift this infrastructure from individual ownership to collective utilization. The big idea is that the ongoing expense of wrangling commodity hardware is better managed by a specialist company. fter all, no carrier or service provider sees any return on this investment of time and treasure and there are significant ongoing costs.
My goal in this article is to define hosted systems and cloud systems, identify the points in common and the points of differentiation between the two and, finally, to discuss why cloud systems for real time applications remains further from becoming a reality than the market (me, included) would like.
Marketing departments have taken great liberties in what they call “cloud” services. Essentially, hosted and managed software, infrastructure and platforms have been around for a while. Outsourcing your hardware and software to a neutral provider (like VoIP Logic) is NOT the same as using cloud systems though there are significant advantages from letting a specialist company host, manage and administrate the systems you use in business operations.
The vision is that enterprise server rooms and even carrier or service provider server farms will be a thing of the past. All of that metal and silicon exists in increasingly vast data centers. In the telecommunications industry—a subset of the much larger data industry—these data centers are devoted primarily to real time communications applications—voice and video—and are instrumental to maintaining the efficacy of communications services.
The growth of hosted systems vs. onsite maintenance of infrastructure has been mirrored in the growth of hosting providers like Equinix, Zayo, Telehouse, Rackspace, etc. In general, both hosted and cloud systems providers meet all of the environmental requirements to maintain carrier-grade infrastructure including redundant and backup power supplies, multi-path network connections, physical and device security as well as cooling, fire protection and 24/7 monitoring and maintenance much more efficiently than any one enterprise, carrier or service provider. cloud and hosted systems diverge when it comes to how the software and hardware behaves.
Using hosted systems is tantamount to physically moving infrastructure from the enterprise or the carrier site and placing it in a shared facility to take advantage of all of the benefits listed above. hosted systems are generally dedicated to a single operator (though partitioning or virtualization of a discrete resource is increasingly possible), are housed in a specific location and do not mitigate the risk of poor performance and multiple points of failure.
Cloud systems, in contrast, do not have a single point of failure, are inherently virtualized, are location independent and do not have scaling constraints. Think of hosted systems as a way station along the path to the real prize—cloud systems.
So, regarding cloud systems for voice and video, are we there yet? Simply put, no. VoIP, SIP, Video and other Real-time Transport Protocols (RTPs) are dependent on physical location and have very specific support requirements. Here is a list of the challenges that need to be overcome to support true cloud systems in telecommunications:
- Ability to capture signaling and media packets is often not available. This is a critical component for voice and video troubleshooting.
- Ability to route and redirect traffic to ancillary systems (like call recording) is complex or impossible to deploy.
- Ability to separate signaling, media and data traffic across multiple network interfaces is a best practice not consistently available in cloud systems.
- Access to network provider quality of service (QoS) mechanisms to ensure quality is not consistently available.
- Specific hardware performing advanced media processing such as DTMF translations, transcoding and RTP detection is not available in a cloud system.
- Root access to systems to program and execute required scripts and small helper applications is generally restricted or impossible.
- IP Addressing flexibility is restricted limiting the ability to divide or subdivide traffic streams into separate realms—crucial for even the smallest operators to manage different classes of service, types of calls, billing, etc.
In conclusion, while cloud systems sound great and play well in advertising and the media, when you pull back the curtain the reality is until more flexible technology is created that allows carriers and service providers to maintain the required flexibility, access and control to which they are accustom (and, it can be argued, is necessary to run their businesses), the concept remains largely a marketing gambit for communications operators.
About the author, Micah Singer, Founder & CEO, VoIPLogic Micah Singer has over 15 years of experience working in telecommunications. Prior to founding VoIP Logic, Mr. Singer's experience included positions as VP, Carrier Relations at Viatel and VP, Business Development at Justice Technology. Mr. Singer has a BA in Political Science and Psychology from Williams College.