Editor's note: This is the second part of a two-part tip series on mobile cloud services. Don't miss part one, Cloud computing and mobile behavior: A new services opportunity
The natural chemistry between mobility and behavior-based services creates a major revenue opportunity, one that's best fulfilled in the cloud due to the dynamic nature of mobile services and spontaneous mobile usage patterns.
But choosing the right technology and features to seize this opportunity is no small feat. It requires strategic, upfront planning about virtualization models, network architecture, management and billing.
The challenge with optimizing cloud architecture for behavior-based mobile cloud services is that the necessary service elements are still evolving. In the past, there was little pressure to run operations support systems (OSS), billing support systems (BSS), network management systems and service delivery platform (SDP) applications on one unified platform. As a result, most operators have a variety of hardware, operating systems (OSes) and middleware elements in place that were used as the early building blocks for creating mobile services.
What's the right virtualization model for mobile cloud services?
It's likely that new service features will have to be created specifically for cloud environments, but this depends on what type of cloud is being supported. The diversity of platforms today seems to argue for an Infrastructure as a Service (IaaS) cloud based on virtualization and supporting multiple guest operating systems, but future needs for flexible application composition also demand service-oriented architecture (SOA).
The challenge with optimizing cloud architecture for behavior-based mobile cloud services is that the necessary service elements are still evolving.
In a sense, IaaS can be regarded as hosted virtualization with two layers of software -- the virtualization software itself, which creates the execution environment for applications, and the cloud software that manages and schedules applications in the resource pool created by virtualization.
There are also two models of virtualization: the hypervisor model of independent virtual machines (VMs) sharing a hardware platform, and the host-OS model in which VMs are run as tasks in a host operating system. These two models differ in how they balance the demands of hosting multiple application platforms against the need to create an optimized SOA framework for the future.
Hypervisor virtualization creates an IaaS cloud that has great flexibility, but it also may present a greater challenge in building behavior-based mobile cloud services. The greatest benefit of the IaaS model lies in the fact that it can cloud-host at least some OSS/BSS components -- that is, the ones compatible with the virtualization platform -- but each VM effectively becomes an independent computer to be managed. Consequently, this model raises operating expenses. In addition, interprocess communication to link components of advanced mobile-cloud services may be harder to coordinate in the hypervisor model because the addresses of the VMs used for application components may change as resources are reassigned.
Although disciplines like DevOps (a mashup of development and operations) are aimed at reducing these problems in the cloud, the hypervisor model of the cloud becomes less useful over time if these new, behavior-based mobile cloud services require a lot of new development work.
In the host-OS model of virtualization, the VMs are a "container" within a normal operating system -- essentially treated as tasks. The multi-tasking capability of the OS then enables it to host applications that are written especially for behavior-based mobile cloud services. The applications and components that need virtualization can then share resources with those that do not.
As a result, the host-OS model is more efficient in running custom cloud-specific applications and components because it doesn't duplicate the OS and middleware as an IaaS-based virtualization solution would. The host-OS model of virtualization, however, achieves a lower degree of isolation among VMs (or between a VM and other applications running on the host OS) than seen in the IaaS model.
Choosing the best network architecture for mobile cloud services
The need to address and isolate mobile application components -- especially those that third parties contribute -- is a significant factor in choosing the best network architecture to support behavior-based mobile cloud services.
Functionally, this kind of cloud is comparable to a content delivery network (CDN), as it must contain interior information paths and an agent through which a user obtains service access. As with CDNs, the architecture to support behavior-based mobile cloud services must provide operators with considerable latitude in building the private network for those services.
Some operators and vendors are considering software-defined networking (SDN) technology like OpenFlow to provide increased Quality of Service (QoS) through improved traffic engineering and to improve security by managing all inter-component flows explicitly. Other providers may want to build their strategy around MPLS or IPv6.
Regardless of the architecture, if the behavior-based mobile cloud services are presented to the user through a cloud portal, then the data, processing and even content needs of those services can be connected to that portal using an optimized private network -- not the open Internet. Developers and partners can be admitted to this trusted interior; of course, the open Internet can also be used for these connections.
Are there too many unresolved technical challenges?
There are still areas where carriers may need to do customization work or use outside professional services.
As noted above, there is an emerging discipline called DevOps designed to treat cloud-service components as application pieces carried in abstract "containers." These containers can then be provisioned onto servers and connected via networks using automated policies. Open-source cloud projects are the main adopters of this approach; their status can be most easily found through an online search using the term DevOps on those projects' websites.
Check out the first part of this tip series
Another evolving issue is the manner of integration between behavior-based mobile cloud services and current OSS/BSS platforms. Although it's likely there will be increased standards interest in this integration, it will be necessary for operators to do some customization work to make those connections inside their back-office systems.
The opportunity for behavior-based mobile cloud services is already here; annual revenues are valued at $500 billion. The technology to support this type of service is mostly available, but even where there are areas that still require special integration and customization, the benefits of taking an early lead in this market could be enormous.
About the author: Tom Nolle is president of CIMI Corporation, a strategic consulting firm specializing in telecommunications and data communications since 1982. He is the publisher of Netwatcher, a journal addressing advanced telecommunications strategy issues. Check out his blog for the latest in communications business and technology development.
This was first published in April 2012