Virtualization technology has long been seen as the core technology of cloud computing, and Infrastructure as a Service (IaaS) has been the cloud service with the widest scope of opportunity. But so far, network operator clouds have focused on hosting services, which means the public face of their offerings is Software as a Service (SaaS) rather than a form of "virtualization hosting"—any offering that lets an enterprise "host" a virtual...
machine on a public platform.
Virtualization is likely to be a good first step in realizing optimal cloud revenues and profits.
Tom Nolle, President, CIMI Corp.
This early history raises the question of what role virtualization will take in the carrier cloud infrastructure and how that role will develop over time. IaaS is a form of virtualization hosting, but some offerings are called "VM hosting," and some enterprises even believe that "dedicated hosting"—meaning buying a public server—qualifies as virtualization hosting.
As a form of hosting, cloud computing is a means of "renting" shared infrastructure to "tenants" who pay only for what they use and thus gain a financial advantage over permanent ownership of computing assets. But from the beginning, there has been a natural tension in the cloud between two competing value issues.
Any X-as-a-Service architecture is most valuable when it offers the most features and displaces the most cost. This suggests that SaaS would be the most valuable cloud service. But SaaS services are applications and therefore have a more narrow scope of value. For example, a sales management application can do only sales management. Lower-level services like Platform as a Service (PaaS) and IaaS have more uses and thus have a broader market.
Virtualization technology provides cloud services a jumping-off point
So how do you balance these forces optimally for revenue and profit? For many operators, virtualization is the answer. A virtual machine is a kind of partition within a server, a logical container into which a "machine image" of operating system, middleware and application software can be loaded. Parallel machine images on the same server are "ships in the night," in that they are as independent of each other as if they were running on their own servers. It is only because they share resources that they have any potential for mutual interference. This model contrasts with operating system facilities for multiprogramming, which normally include some ability to communicate between applications that are running and some potential to share data resources.
For network operators looking at cloud computing, virtualization is a logical starting point because virtualized data centers can be used to support any software platform and application set that can be run on the native machines such as x86. Here are some ways that virtualization technology can help with cloud services:
- Facilities can be used to host customer machine images and thus be sold directly as an IaaS cloud computing service similar to Amazon's EC2.
- By adding an operating system and middleware to these virtual machines, operators can offer PaaS in a way that mimics what Microsoft delivers with its Azure cloud service.
- Finally, if the whole software stack is provided, including applications, the virtual platform hosts SaaS.
This adds up to an interesting point for the operator cloud, which is that no matter what level of cloud service (IaaS through SaaS) operators might plan to offer and no matter where their revenue priorities lie in those levels, virtualization is likely to be a good first step in realizing optimal cloud revenues and profits.
Virtualization technology drives cloud infrastructure choices
Virtualization technology also creates some interesting possibilities in terms of infrastructure evolution.
x86 and blade servers. Nearly all lower-layer cloud services, and the most popular virtualization technologies, are based on the x86 platform. That means that the decision to base operator cloud infrastructure on virtualization technology would likely pull the operator further toward x86 technology over time, something that could reduce the opportunity for other processor types that aren't object-code-compatible with the x86. The blade server architecture is also the normal framework for deploying virtualization, so blade servers are likely the form factor on which this growing x86 interest would focus.
Open source adoption and OSS/BSS. Most virtualization and cloud technology tools are open source, and network operators have historically not been leading-edge players in open source adoption. This is particularly true for their own OSS/BSS or internal IT functions. Yet there are indications that the decision to build x86 and data centers primarily made up of servers running virtualization for the cloud is creating greater interest in both of these technologies for service providers' OSS/BSS functionality. While the numerical majority of operators worldwide already use x86 for their OSS/BSS platforms, most have relatively few open source OSS/BSS components, and that may now change.
Service feature hosting. The virtualization trend also opens the possibility that cloud infrastructure could begin to absorb service feature hosting applications that were once created on service delivery platforms (SDPs) Many operators report that they have deployed thousands or even tens of thousands of SDPs that are significantly underutilized. That is an eerie mirror of the situation that enterprises faced with server proliferation—which was the primary driver for consolidation and virtualization in the enterprise space.
Increasingly, operators see SDP roles limited to applications that require server technology deployed in a central office. These roles are also limited to applications that require Network Equipment-Building System (NEBS) compliance—the set of safety, spatial and environmental design guidelines for telecommunications equipment in the United States—or in voice-feature-related applications linked to IMS.
Virtualization technology offers service providers cloud flexibility
Service providers still have not fully committed to a single cloud strategy, much less a strategy for hosting cloud services. But they are committed to offering cloud services and deploying elements of their OSS/BSS systems and even service features in the cloud. Virtualization technology offers more flexibility in matching variable customer cloud needs to a static set of cloud servers. That means that virtualization is a good way to hedge early bets in infrastructure spending and to maximize flexibility when the market for cloud services and the sweet spot for network operators in that market have yet to be determined.
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 SearchTelecom.com networking blog, Uncommon Wisdom.