Read 70326_comarch text version


AMD OpteronTM Processors Deliver Outstanding Performance for Telecommunications Business Support Systems (BSS)


Executive Summary

Increasing challenges in billing systems demand reliable and scalable standard computing server technology with increasing performance to handle the increasing complexity and number of subscriber, service, and pricing data. COMARCH has tested a Dual-Core AMD OpteronTM processor-based server with its Billing System. Wherever there is a need for processing of large amounts of data such as rating of telecommunications subscriber usage data or generating detailed invoices for millions of subscribers, Dual-Core AMD Opteron processors can be relied upon to deliver outstanding performance resulting in the processing of more billing requests at a time with lower power consumption and space requirements, providing higher efficiency to communication service providers. The following test report will introduce the reader to the challenges of billing systems as part of business support systems; COMARCH's billing system and capabilities of AMD Opteron processors. A method to evaluate real world performance of billing systems is given and test results are listed.


Business Support Systems (BSS) in Telecommunications

Business Support Systems (BSS) are used by telecommunication operators to run business operations. Business Support Systems cover four main areas: product management, customer management, revenue management, and fulfillment management. Typical activities of Business Support Systems are: taking customer orders, managing customer data, managing order data, billing, rating, and offering business to business (B2B) and business to customer (B2C) services. The enhanced Telecom Operations Map (eTOM) delivers a business process model and describes all the enterprise processes required by a service provider.

AMD64 Processors for Telecom Business Support Systems 2.1 Issues and challenges with revenue management

With the move of service providers to next generation network architectures and convergent services, the challenges in the area of revenue management increase. With the increasing number of service components and product offerings, increasing number of billing options, and a greater choice of payment options that are offered to customers, the amount and complexity of data to manage increases exponentially. On the service provider side, the pressure for higher efficiency of BSS systems increases. Last but not least, the exceptional expectations for reliability of billing systems in telecommunications, is at the heart of revenue collection.


COMARCH Billing System

COMARCH Billing System is a comprehensive response to the main concerns of next generation telecommunications providers. It is a high capacity, fully scalable system with a powerful processing engine that rapidly handles virtually unlimited amounts of data. It has been designed to support carriers by offering convergent next generation services thanks to event-driven 3G billing with flexible event format definitions for IP, VoIP, GPRS and UMTS, including a wide range of tariffs and discounting schemes.

Figure 1

COMARCH Billing System Tariff Plan Managemenet

The system is dedicated to all communication service providers regardless of their size, traffic, services or ambitions; from small start-up operators to well-established carriers. It enables communication providers to optimize their next generation business strategies, networks and services, thereby changing the way they conduct business. COMARCH Billing System can benefit operators by maximizing revenue (i.e. by improving customer service, decreasing churn), implementing totally new clearing schemes, extending the offer of various discounts, from very simple to highly sophisticated (i.e. depending on customer usage), assessing charges based on multiple factors (i.e. time, bytes, event content at any given time), and maintaining a unique competitive advantage over competitors by delivering technologically advanced software solutions. 2

AMD64 Processors for Telecom Business Support Systems At the heart of the COMARCH Billing System resides DPS, Data Processing Server. State-of-the-art advancements in IT technologies and many years of experience in the construction and implementation of efficient mass data processing mechanisms were employed by COMARCH computer scientists while designing this highly efficient, internally developed, flexible engine used for mass data processing. This multi-threaded system works in the UNIX/Linux environment and allows for parallel processing of any event data. Deployment of DPS results in significant reduction of time needed for performing billing processes such as usage record loading, rating, invoicing, or payment collection and clearing. Efficiency of the system makes it possible to process even the most complex tariff definitions quickly and accurately.

Figure 2

COMARCH Billing System Components

In a system capable of billing millions of telecommunications subscribers and processing hundreds of millions of usage records per day, the hardware component plays an extremely important and integral role. While COMARCH Billing System is platform independent, COMARCH is always on the lookout for cost-effective and efficient hardware solutions, which can be proposed to the telecom operators in order to meet their existing and future billing needs. After all, the speed of mass processing of billing events depends not only on the advanced software architecture that COMARCH created, but also on the machines on which it is installed. Dual-Core AMD Opteron processors have proven to be an extremely competitive and cost-effective alternative to other mainstream CPU offerings using legacy Front-Side-Bus Architectures.


AMD64 based Computing Platforms for Telecommunications

From a hardware point of view, key equipment within Next-Generation-Networks is based on server computing technology. Beyond standard server products for NextGeneration-Networks, AMD Opteron processors are also available in open standard chassis platforms like AdvancedTCA (ATCA), BladeCenter and CompactPCI. These platforms are building blocks for solutions that provide the longevity and reliability required by service providers in communication networks such as IP Multimedia Subsystem (IMS), Voice over IP (VoIP), Global System for Mobile Communications 3

AMD64 Processors for Telecom Business Support Systems (GSM), and Code Division Multiple Access (CDMA) wireless networks. At this level, AMD64 technology provides several unique customer value propositions: 4.1 Industry-Leading Performance-Per-Watt

With the equipment changes and upgrades to accommodate new services and additional throughput, the costs from power consumption and air conditioning are becoming a major concern for service providers. AMD PowerNow!TM technology with Optimized Power Management provides power on demand computing, and can reduce CPU power consumption by as much as 43 percent while under an approximate 60 percent load, or up to 75% reduction in CPU power when idle. Adjusting CPU power for the load creates a ripple effect of benefits in the data center that can increase dramatically as the number of processors increases. These benefits can include reduction in the amount of heat produced by the processor and other components, and reduction in the amount of electricity required to both run and cool the system. By offering a processor that provides for a dense compute configuration with low overall power consumption, AMD helps make it possible to put more compute capability in data centers and central offices and potentially lower utility costs. 4.2 Scalability

Evolutionary deployment of new technology like IMS and 3G must be based on scalable architectures and the long-term protection of past investments. To this end, AMD64 technology offers an innovative solution that allows network operators and service providers to run 32-bit and 64-bit applications simultaneously ­ operators and service providers can grow into 64-bit computing without sacrificing their existing investments. Perhaps even more importantly, AMD64 technology was designed from the ground up with multi-core computing in mind, resulting in high performance for multi-threaded and multi-tasking environments. Quad-Core AMD Opteron processors will be socket compatible with today's DDR2-based AMD Opteron processors, offering a seamless upgrade path to quad-core performance as demand increases. This simplified platform approach to implementing multi-core technology is further enhanced by Direct Connect Architecture, which provides exceptional system performance and multi-processor scalability in up to 8-processor systems. 4.3 AMD's revolutionary Direct Connect Architecture

AMD64 technology with Direct Connect Architecture was designed to improve overall system performance and efficiency by directly connecting the processors, the memory, and the I/O. Directly connecting these elements helps eliminate the front-side bus bottleneck inherent in legacy x86 architectures.


AMD64 Processors for Telecom Business Support Systems On the AMD Opteron processor, the integrated DDR2 DRAM memory controller changes the way the processor accesses main memory to increase bandwidth and reduce memory latencies. With dual channel RDDR2-667 available on the AMD Opteron 2000 and 8000 Series processors and UDDR2-800 available on the AMD Opteron 1000 Series processors, theoretical maximum memory bandwidth (per core) is between 10.6GB/s and 12.8GB/s. Another benefit of integrating the memory controller directly on the processor is that memory bandwidth scales with the addition of processors. This helps provide highthroughput responsiveness and scalability with multi-processor platforms. AMD64 technology with Direct Connect Architecture features HyperTransportTM technology and provides up to 8.0GB/s bandwidth per HyperTransport link-- sufficient bandwidth for interconnect technologies including PCI Express®, PCI-X®, DDR, InfiniBand, and 10G Ethernet. AMD Opteron processors support up to three HyperTransport links per processor, providing up to 24.0GB/s peak bandwidth per processor. These processor-level features can provide a tremendous competitive advantage for database applications. AMD's Direct Connect Architecture allows each processor in the multi-processor compute node to access the memory of the other processors at low latency with high bandwidth. The up to 256 terabytes of memory address space of AMD Opteron processors helps the performance of applications when large (or many) datasets are held in memory, as is the case with billing systems. 4.4 AMD VirtualizationTM Technology (AMD-VTM)

Second-Generation AMD Opteron processors are designed to help end the "one server, one application" problem. With hardware-assisted AMD VirtualizationTM technology (AMD-VTM), Second-Generation AMD Opteron processors can help streamline the Telecom infrastructure to achieve higher levels of efficiency and utilization. Regardless of the specific network implementation chosen by a particular service provider, AMD-VTM helps enable a flexible, one­to-many relationship between hardware, operating systems, and applications, and contributes to functional flexibility, implementation, and distribution for network equipment vendors.



Testing System Performance with COMARCH Billing System

Introduction: the rating process

In order to determine the performance of the Second-Generation AMD Opteron processors in a real-life application, COMARCH decided on having its Billing System perform rating. The process of rating of call detail records (CDR) constitutes a key element of the entire billing lifecycle, and its efficiency is often used to differentiate COMARCH from its competitors. In simple terms, the process of rating entails analyzing of CDRs generated by network elements (for example, voice switches) and calculating charges based on tariff plans defined in the billing system and allocating them to the account of a given subscriber. Typically, the charges may depend on a number of 5

AMD64 Processors for Telecom Business Support Systems factors including: call duration (number of minutes or seconds), time band (day, night, weekday, weekend), and destination (local, long distance, international, fixed, mobile). The rating process consists of four steps: 1. LOADING ­ loading stage handles acceptance of the input data and transformation into internal structures used by the DPS module and COMARCH Billing System database; 2. GUIDING ­ finding the appropriate contracting party in the COMARCH Billing System database, who should be associated with the incoming records and charges; 3. PREFIX ANALYSIS ­ finding the prefix most suitable for the fields containing information about the calling party and the called party. Prefix analysis helps determine geographic location or service performed. It also permits the calculation of distance between two geographic points, should they be given or computed from the geographic coordinate prefixes; 4. PRICING PROGRAM ­ a structure containing the successive stages of computations: analyzing contract items, account hierarchy, product dependencies, rule-based event and charge filtering, fee aggregations 5.2 Test environment

Tests were performed on an 2P AMD Opteron processor based system.



2x AMD Opteron Model 2218

Table 1

CPU Speed

2.6 GHz per proc


Mem Total [kB]

4 GByte DDR2-667

Tyan S3992


Hardware configuration of system under test

Operating System

RHES4U4 ­ x86_64

Oracle Version

Billing System Version


DPS Version for Linux x86_64

6.4.3-857 x86_64

Table 2

Software configuration of system under test


AMD64 Processors for Telecom Business Support Systems 5.3 Test scenarios and results

For test purposes, COMARCH Billing System processed call records from raw CDR files and stored this data in an Oracle database. In order to measure CDR processing times, two series of tests were carried out: 1. Rating and storing data in database 2. Rating without storing data in database 5.3.1 Scenario 1

In this scenario, COMARCH Billing System performed rating of 10 files with a total of 365,000 records. The results were then stored in the database. This type of test reflects a typical operation of the system where large numbers of records are rated daily and stored for future use in the invoicing process. The test was repeated three times for a different number of software threads. Results are provided in records per second.

# of Threads

Records processed [/sec]

Table 3







Test results for test scenario 1

5.3.2 Scenario 2 In preparation of the test in the second scenario, a small file with 307 records was initially processed and stored in the database. Then, during the actual test, a shell script was executed to re-rate these records 100 times (loop), without storing the results in database. This type of test helps isolate CPU performance from the performance of other hardware components such as disks or channels. Each test was repeated 10 times and the results averaged in order to improve accuracy. Results are provided in records per second.

# of Threads

Records processed [/sec]

Table 4









Test results for test scenario 2




At a very competitive price point, Second-Generation AMD Opteron processors deliver a great deal of computing power which is required by today's resource intensive back office applications such as COMARCH Billing System. Wherever there is a need for processing of large amounts of data such as rating of telecommunications subscriber usage data or generating detailed invoices for millions of subscribers, these CPUs can be relied upon to deliver outstanding performance. AMD has proven to be a very competitive player not just in the realm of home PCs but also in the world of large enterprise systems where performance, scalability, and reliability translate into business success.

Additional resources

AMD Opteron processors -

Comarch OSS/BSS ­ Comarch, Inc. 1320 Tower Rd Schaumburg, Illinois 60173, USA Phone 800 786 4408 Fax 800 684 5916

Advanced Micro Devices One AMD Place P.O. Box 3453 Sunnyvale, CA 94088-3453 ©Advanced Micro Devices 2007. All rights reserved. AMD, the AMD arrow logo, AMD Opteron, and combinations thereof, AMD PowerNow!, AMD Virtualization, and AMD-V are trademarks of Advanced Micro Devices, Inc. HyperTransport is a licensed trademark of the HyperTransport Technology Consortium. PCI-X and PCI Express are registered trademarks of PCI_SIG. Linux is a registered trademark of Linus Torvalds. Other names used in this document are for identification purposes only and may be trademarks of their respective companies. 43283-A



8 pages

Find more like this

Report File (DMCA)

Our content is added by our users. We aim to remove reported files within 1 working day. Please use this link to notify us:

Report this file as copyright or inappropriate


You might also be interested in

Single Cover