IP Mediation: System Selection and Functionality Requirements

Mediation is considered to be the lynchpin for usage-based services. TeleStrategies recently released the 2001 IP Mediation Functionality Report, which highlights the differences in functionality supported by the various players. The results demonstrate that not all mediation systems are the same, nor do they all target the same marketplace. The vendors participating in the report were tested and evaluated across each of these areas: Configuration versus customization, services and protocols supported, convergent services, collection capabilities, database archival and storage functionality, provisioning, and performance. (See sidebar, “Participants in 2001 IP Mediation Functionality Report.”)

Configuration Versus Customization

Systems differentiated themselves mainly by their GUIs and configurable components, not by architecture. In general, they feature object-oriented programming design, which allows linking modules of functionality to suit the customer’s requirements and business logic.

And just what is the right mix of configuration and customization? Configuration with some flexibility for extension is the ideal model for a mediation system. Given the immaturity of the mediation market, the fact is that customization, although it allows for considerable expansion, takes too long. Providers would rather implement a system that can be configured, even though the trade-off is having to trust that the vendor has defined the broadest set of choices.

Each system’s architecture involves components that collect data from network devices; perform elements of aggregation such as filtering for duplicates and erroneous files; reformat the data to an internal structure; further aggregate, supplement and correlate the data into records; and then translate the records to various formats to be sent downstream to billing and other OSS components. Of the 15 systems reviewed, all featured a multi-tiered architecture that separated collection from data operations and from distribution to other systems. They typically manipulated data by first translating it into a normalized, often proprietary, format to then begin the work of applying aggregation operations to the data.

An IP mediation system with the ideal mix of functionality would be able to collect data using multiple protocols, and would have filtering and error detection as configurable as possible, extendable aggregation, and ASCII-formatted archival functionality.

About 67 percent of vendors provide some kind of software development toolkit for enhancements to the system. Of those, only a few provide either an internal SQL-type programming language or a development environment at no additional cost. As for integration, again about 67 percent of vendors provide some form of APIs for the system. Most common are those for OSS and billing components.

However, providing SDK-type functionality is critical, whether at an additional price or not, because it is the only way to create new business rules, modify mediation rules, filter data and handle other such modifications, and is considered the most critical factor for ensuring a carriers’ success.

Services and Protocols Supported

Vendors were evaluated on which services they support. All vendors claim to support VoIP services, AAA services, and VPNs. Content provider services and wholesale services tied for second (93 percent) as the most commonly supported.

Carriers should be looking to implement applications that either support multiple protocols or can provide the flexibility to add them with relative ease. For example, providing Internet services, network transport and video streaming services would require a collection application that supports HTTP, SNMP and RSVP protocols, respectively. In addition, the carrier might use NetFlow to analyze network traffic trends and to plan capacity.

Comptel, Intec Telecom Systems, Narus and Xacct support the largest number of protocols, which in turn allow for the largest number of network configurations.

Convergent Services

Vendors were asked if their packages were designed to provide convergent services for packet- and circuit-based mediation. Nearly three-fourths (73 percent) responded that their product was designed to support both IP and traditional voice mediation. Most carriers have spent large sums implementing collection and mediation applications to support their voice switches. In addition, they have invested even more to integrate these applications with downstream applications, and they are in no hurry to rip that functionality out. Support for both voice as well as IP is an added benefit, but not necessarily a practical one. Again, the critical factor in the mediation system’s success is its ability to be integrated into the provider’s current network environment.


Most systems are marketed as “plug and play” in relation to integration with downstream applications, but significant time is expended in configuring the adaptors to get the integration working. The best applications are those that have the most experience with existing commercial billing applications. However, carriers must also consider how much the commercial billing system has been customized. If there has been heavy customization, intermediate agents may need to be developed to export data from the mediation archive media and transform the data into a common input format acceptable to the billing application. This way, the mediation application can be extended without voiding the warranty, affecting future product upgrades or compromising product support.

Although several mediation vendors have partnered with billing vendors, integration between the applications is still complex. If adaptors are prepackaged to support the integration of partner applications, then the issue becomes one of configuration. However, if not prepackaged, then it depends on the billing application used by the carrier and whether it provides an API to feed the mediated data to pricing. Lastly and more important, the ability to correlate usage to the customer account becomes the key factor for the success of both mediation and billing.

Collection Mechanisms

Vendors were assessed on how the system collects data from devices. With a few exceptions, all vendors support a mixture of monitoring, probing and event-based collection that is highly dependent on both the source and functional range of the services being collected.

Though defining the pros and cons of collection methods is tricky, the key factor is whether the carrier wants the data in near real time versus receiving log files periodically. The IP mediation system should support both approaches, though it is more important that any collection approach does not degrade the network due to additional traffic overhead.

All of the vendors were tested on support for data collection by batch process and in real time. Definitions of “real time” varied; though systems can stand up to the real-time test, this aspect should be configurable, meaning that it is up to the carrier to decide which mode to implement.

Vendors were also asked if the mediation system could collect usage data for GPRS from multiple mobile switching centers. Nearly three-fourths of the vendors (74 percent) indicated that they support this type of collection. The advancement of support for GPRS multiple mobile switching centers is questionable, however, as implementation is still in its infancy.

Database Support

and Architecture

Two-thirds of the vendors (67 percent) are partnering with Oracle. No surprise there; however, no one is natively using a commercial relational database management system to do the actual work of aggregation and correlation. Most vendors stated that the frontline data crunching is done through internal processing and persistent memory storage. The most reliable method is still to use flat file storage and then push aggregated records out to a data warehouse for access by other systems, including billing, CRM and fraud.

Database functionality for storage and manipulation of data is critical, including the ability to store data in other formats such as ASCII files. Further, the ability to use database tables to store billable records and flat files to store data for traffic analysis is an economical decision due to the high cost of database storage for non-billable data.

However, the ability of carriers to massage data, after initial collection and filtering, is not about economics—it is one of the most essential functionality points of mediation. Without it the product becomes a collection application only. Data must be massaged before being fed to the rating engine, unwanted data has to be filtered, and other types of data must be added for trend analysis, alarm monitoring, and the like. So if the carrier cannot massage the data, then there’s no point in procuring such a system.

Provisioning

Vendors were essentially split down the middle in terms of support for provisioning functionality for prepaid usage and the ability to perform mid-call teardown operations. More than half of the vendors (57 percent) do not currently support provisioning; however, 14 percent of those can offer the functionality with the purchase of additional components or are developing provisioning solutions.

Though provisioning is not a necessary component of the mediation system, mediation must be considered during the provisioning cycle, because it is here that a specific circuit is assigned to the customer account. Data collected and mediated must have a specific identifier that will allow downstream applications to correlate the traffic to a customer account.

Performance and Scalability

Vendors were generally unable to provide formal benchmark data. Only 20 percent of the vendors could say they had participated in benchmarking scenarios where their customer had been the evaluator. The remaining 80 percent either had not conducted any form of benchmarking, or had internal or non-publishable results.

It is not surprising that performance testing is so infrequent, given the immaturity and complexity of the market space. Benchmarking tests will probably not be released anytime soon, because scalability will be a major determining factor for which systems survive and which ones fail.

Currently, mediation vendors are not exploring all the extensive metrics data provided by network elements. As an example, most of them do not provide the ability to audit network elements for true network latency, even though the survivors in the market will want to do so. Currently, carriers must buy additional products, such as Trendium or Externa, to obtain metrics that focus on collecting exceptions from the network elements like cell error ratio, network latency, data delivery rate, mean cell rate, and unavailable seconds instead of obtaining metrics from collected and mediated network usage such as bytes in/out.

Market Presence

The report’s sections on market presence and future targets provide insight into the IP mediation market’s current and expected focus. About three-fourths of the vendors (73 percent) include wireless in their current market base. In addition, 64 percent are targeting the wireless market within the next year. Of the latter, 44 percent will be attempting to break into wireless for the first time.

Though 60 percent of vendors also claim CLECs as part of their current market presence, none are targeting CLECs within the next year. ILECs and business/backbone ISP providers remain favorable market targets, with 36 percent and 21 percent of vendors, respectively, looking to break into those markets. How do market share and focus match up with customer numbers? See the table “Current and Projected Installations.”

Pricing and Integration

Vendors were asked for information about pricing—how they priced and licensed their systems, what percentage of the total system cost included service, what percentage of the total system cost included integration, and the time frame for integration.

Forty-seven percent of the vendors set prices based on per-customer negotiation as well as on published price lists. Twenty-seven percent provide negotiated pricing in every case, and 20 percent provide a published price list in each case. Most of the 47 percent who do both indicated that pricing is negotiated generally for large customers only.

For medium-sized customers, 24 percent of vendors claim service costs are up to 10 percent of first-year costs, 38 percent estimate service costs to be between 10 and 25 percent, and 38 percent estimate service costs to be between 25 and 50 percent of the total first-year costs. The same breakdown applies to estimates for integration for a medium-sized customer (24 percent reported less than 10 percent integration costs, 38 percent reported between 10 and 25 percent, and 38 percent estimated between 25 and 50 percent of total first-year costs). Thirteen percent of the vendors did not respond to the pricing section of the survey.

Estimated times for installation and integration veer to the side of optimism, with most vendors indicating ranges of a few weeks to about three months. For larger installations, Intec Telecom Systems and Xacct Technologies realistically stated that it can take up to nine months. Vendors were also quick to list such factors as system complexity and network component readiness as affecting installation and integration time.

For more information on the 2001 IP Mediation Functionality Report, contact TeleStrategies at 703.734.7050 or reports@telestrategies.com.

Stacy Hunter is senior reports editor at TeleStrategies, Inc. Her professional background is in software and systems engineering for large-scale communications systems and secure Internet-based applications.

Julio Prato is chief operating officer and consultant at Frederick Hamilton Rhys Management Group LLC. He can be reached at jprato@fhrmg.com.