The hard facts about Softswitches

Service providers know that the Internet represents a tremendous revenue-generating opportunity. They also are learning that finding the best way to tap into that revenue stream can be tricky at best.

Softswitches are a case in point. This class of equipment uses software intelligence to allow carriers to integrate voice and multimedia services over existing ATM or time division multiplexing (TDM) networks or the Internet. The primary value of softswitches is they allow service providers to quickly roll out new services without involving switch vendors in the process. That is why softswitches create such promising opportunities for service providers to quickly ramp up their Internet revenues.

But here is the tricky part: Picking the right softswitch is hard work. Not only can providers choose from a huge number of products - more than 70 vendors have joined the Softswitch Consortium - but no standard definition for a softswitch exists. Vendors themselves have almost as many definitions as they do products.

Within that vacuum, service providers create their own working softswitch definitions based on what they need the softswitch to do, then they matchtheir definition with available products. However, some items should be on every service provider's softswitch wish list, including:

- Support for service creation

- An open, directory-enabled platform

- A secure broadband call model

- Bulletproof reliability

- Straightforward integration with legacy networks

- Scalable intelligence and call processing

- Support for class-independent services over any media

The creation process

Typically, softswitches are installed between TDM and ATM phone networks and an IP packet data network to provide the necessary intelligence and signaling smarts to route phone calls between the two. Softswitches address service provider pressures to improve service delivery while reducing costs. Pivotal to integration success is a comprehensive solution that not only offers innovative services for the service-enabled IP infrastructure but also allows mediation with existing circuit-switched technology.

Service creation is therefore the obvious place to start a detailed evaluation of softswitches. It is important to ask whether the softswitch provides a platform and architecture that is flexible and open enough to accommodate a variety of value-added services.

Ironically, the important thing here is not to get bogged down in spec sheet details such as whether a product supports 1-800 features or a specific service such as local number portability.

Instead, it is better to look at the design to see whether the softswitch is open and extensible, so that it will support not only the current crop of revenue-generating, value-added services, but also services that don't exist yet.

How is a carrier to tell? It should look for standard signaling protocol support - such as SS7, H.323, MGCP and SIP - as well as open application programming interfaces (APIs) such as JAIN and Parlay. The softswitch should support legacy and next generation interfaces in a seamless integration of traditional and "new world" technologies.

Softswitches should enable a service creation environment that puts service providers in control by allowing a choice of turnkey and mix-and-match solutions. Softswitches can enable converged service creation environments with increased feature sets by integrating legacy and next generation interfaces. By providing control over IP, ATM and legacy TDM connections through an open systems platform, softswitches can provide any-to-any service mediation. By leveraging the latest in directory-enabled technology, softswitches can provide consistent subscriber profile management and allow providers to offer measurable service level agreements (Figure 1).

Customizable switches

The public network architecture is reliable and secure, but circuit switches will not evolve to broadband call models. The Internet architecture is flexible and fast, but hardware redundancy and best-effort service is not enough for non-stop networks. Therefore, it is important that softswitches are based on broadband call models that enable high levels of IP security to protect against denial of service attacks and to provide firewall control for IP applications. Other important functions call models must perform include call signaling and processing, call admission, authorization and authentication, address translation and zone management, usage recording and feature processing.

It also is important that softswitch platforms allow end users to customize their own voice services over the Internet quickly and easily. Some softswitch vendors are using Java-based, directory-enabled software to do this. These products allow end users to accessorize basic voice service with everything from virtual private network profiles to voice mail - all from a simple Web browser interface.

This is particularly important because rapid time to revenue requires the swift development of innovative applications and the accelerated provisioning of new services. Providers need a robust, easy-to-use interface for fast and consistent rollouts of new services. Empowering the customer to easily embellish or accessorize basic voice service increases revenue while reducing the cost of supporting the service (Figure 2). It also cuts down on churn by making it less likely that customers will want to switch services providers and risk losing key features and flexible service configurations.

By enabling Web-based self-service, providers can improve customer satisfaction while accelerating revenue recognition, creating a competitive barrier to entry and optimizing operational productivity.

Affordable reliability

Of course, it doesn't matter how much money the service provider can make from these value-added bells and whistles if the softswitch itself is not reliable. One approach to ensuring five-nines reliability is to run the softswitch software on highly redundant, high-end servers. The snag is that these servers cost around $400,000. Worse, expensive hardware only addresses half the problem.

To ensure maximum uptime, the software that runs on the server also must be protected. A promising solution gaining popularity is to use clustering software that can spread the softswitch intelligence over multiple servers. This approach increases the reliability of the network because clusters are less vulnerable to software bugs. The load-balancing features inherent in clustering applications also improve performance by allowing processing tasks to be spread over multiple servers. Clustering also reduces capital outlay because it allows less expensive server hardware to be installed for between $30,000 and $40,000.

The ability to fit into a service provider's installed network with a minimum of disruption is another key. Most softswitch solutions are designed to link IP, ATM or TDM networks over an IP backbone. But that assumes that every part of a service provider's network is built around a core IP network (Figure 3).

However, this is not always the case in today's real-world networks. For example, consider a global carrier that already has built a solid ATM backbone in Geneva, a highly profitable TDM backbone in Karachi, Pakistan, and a relatively new IP backbone in Los Angeles. For now, this carrier needs the flexibility to use softswitches to connect to its IP backbone in the U.S. and wants the flexibility to use the other backbone networks where appropriate. That means finding a product that delivers any-to-any connectivity by hooking up any combination of IP, ATM and TDM over a backbone built using any combination of these technologies. This way, service providers can map a migration path to IP without having to make their existing - and profitable - service infrastructure obsolete.

Scalability is also a must in any softswitch definition. The best way to measure the scalability of a softswitch is to look at its underlying architecture. Attaching old networks to new networks actually requires two basic components: the softswitch itself, which provides the network intelligence, and a media gateway, which provides the physical interfaces between the networks and performs the necessary protocol conversions.

Some vendors separate these components into completely different devices. But another category lets service providers choose where to put the intelligence by providing stand-alone softswitches that can be used with media gateways, which also support some distributed intelligence. This provides improved scalability by allowing service providers to offload certain processing tasks - such as call blocking - to a media gateway to accelerate performance in larger networks.

Softswitches that deliver full compatibility with voice and data environments should be able to perform call agent functions to MGCP clouds and H.323 gatekeeper functions in H.323 environments, while also delivering native IP services such as bandwidth and firewall management of LAN/WAN devices. The distance-independent nature of IP allows the scaling of geographically disparate multishelf/multipoint of presence systems to "virtual switches." This flexibility forces a shift from traditional Class 5/Class 4 services to allow class-independent services over any media.

Switching's future has arrived

Network deployments now can become class-independent with services residing as close to the service consumer as possible. The service logic can be kept in a central location, distributed to the edge of the network or scripted to the media gateway - independent of whether the last-mile technology is IP, ATM or even legacy TDM.

By delivering a flexible call engine with extensive APIs on top of basic call services, softswitches can enable the ultimate flexibility in service deployment solutions. This includes uncomplicated integration with third-party applications via protocol or programming interfaces, and meta-directory technology for back-office integration. Softswitches allow rapid service prototyping and next generation service creation environments that leverage the latest in IP technology.

There is no doubt the softswitch category requires a lot of thought and consideration. But it also is increasingly clear that service providers cannot afford to ignore this product class. Creating its own softswitch definition and matching it to the available products is the fastest way for a service provider to identify the product that will enable it to offer new and varied voice service quickly and easily.