News Archive (1999-2012) | 2013-current at LinuxGizmos | Current Tech News Portal |    About   

Whitepaper: The benefits of modular communications platforms

Feb 10, 2005 — by LinuxDevices Staff — from the LinuxDevices Archive — 2 views

Foreword: Economic pressures are driving telecom equipment manufacturers to abandon proprietary platforms in favor of modular, standardized architectures that reduces costs and development time, according to a new Intel whitepaper. The paper reviews Intel's efforts to standardize multiple levels of telecom infrastructure, with technologies such as AdvancedTCA and Carrier Grade Linux.


Promoting the Benefits of Modular Communications Platforms

by Aniruddha Kundu and Udayan Mukherjee


Overview: Benefits of Standardization

The telecommunications industry is facing increasingly competitive pressure to support more users and increase the variety, reliability and availability of services, all while lowering costs. To meet these pressures, telecommunications providers are looking to telecomm equipment manufacturers (TEMs) for lower cost, more scalable solutions. Intel recognized these needs several years ago, and has been working on standards efforts at all levels of telecommunications infrastructure to promote interoperability, lower costs and shorten time to market for telecommunications platforms.

Timing is everything. In October of 2001, three Intel employees flew to Chicago to present an idea for a next-generation standard for telecommunications architecture to the PCI Industrial Computer Manufacturers Group, or PICMG. The telecommunications industry was still using mainly proprietary platforms that took years to design and build. Many telecommunications equipment providers feared that component standardization would cause their proprietary platforms to lose value, and Intel's proposals were not particularly welcomed.

Today, increasing economic pressures in the telecommunications industry have proven the need for standardization, and Intel is actively involved in standardizing hardware and software at every level of telecommunications systems architecture. These efforts are helping to create interoperability and a rich ecosystem of modular communications components that will lower costs and time to market for original equipment manufacturers (OEMs) and TEMs.

From Boom to Trust

In the late 1990s, there were numerous telecommunications platforms available from TEMs and OEMS, but few standards and no interoperability. By 2001, the only telecommunications architecture standard on the horizon was the PICMG 2.0 family of specifications. Developed by a group of proprietary systems vendors, the standard was vague, so if vendors claimed to be “standards-based,” there might still be no interoperability. Proprietary telecommunications platforms were expensive and time-consuming to build, with development cycles of two to four years, and that cost was passed along to customers. But the Internet economy and the telecommunications industry had been booming, and telecommunications service providers (TSPs) had been willing and able to pay big money for stand-alone, proprietary platforms.

But big changes were brewing. In 1999, John Miner of Intel's Desktop Platforms Group started a communications platform group (CPG). At the time, the proposed platform was for the data center market. The goal was to create a standard architecture to enable interoperability between platforms. The platform architecture had five major goals:

  • Enable interoperability
  • Reduce time to market
  • Reduce development costs
  • Reduce the length of development cycles
  • Create an open, standards-based scalable, flexible platform architecture

By late 2001, the group had determined that its modular platform architecture could potentially reduce development time and costs by at least 30 percent, and Intel decided it was time to present a proposal to a standards body. In October of 2001, Intel presented to PICMG the idea of developing a next-generation standard for communications platforms. Some members of PICMG saw the benefits of interoperability and were in favor of the idea. Others were against the Intel proposal because the PICMG 2.16 standard had just been defined, and some vendors were afraid their proprietary platforms could lose value with standardization.

Intel, however, persisted in its discussions with the standards bodies, and as the dot-com boom ended and the telecommunications industry began to slump, both platform vendors and TSPs faced increased competition and lower profit margins. Infrastructure spending slowed. At the same time, service providers were pushing TEMs and OEMS for more scalability and system features to support more users and new applications such as Voice over IP (VoIP), text messaging, and even video-on-demand (VOD).

Telecommunications equipment makers became more open to standards, realizing that they didn't have enough engineering talent or resources to continue spending years developing their proprietary platforms and meet their customers' demands for lower prices and increasing capabilities. Through all these changes, Intel kept up a dialog with those vendors, repeatedly pointing out that platforms are more than hardware and that the real value-add is in applications software, manageability, security, availability, and other aspects of the platform.

Intel's Standards Efforts

Today a new vision is emerging in the communications space, similar to the evolution of the compute server industry a decade ago. Ten years ago, servers were monolithic, proprietary systems. Today, you can combine totally modular boards, boxes, and software, put them together, and innovate on top of that open platform. Intel is working in three main areas to help drive standards for an equally modular communications computing architecture.

Out of Intel's initial efforts with PICMG evolved the Advanced Telecom Computing Architecture (AdvancedTCA) standard, a robust (in fact, currently the only) standard architecture for bladed telecommunications servers. AdvancedTCA specifies form factors for backplanes, boards, and so forth. But to make a complete telecommunications platform, telecomm applications also need a carrier-grade operating system and carrier-grade middleware. So in addition to working with PICMG, Intel began efforts with two other standards bodies, the Service Availability* Forum and the Open Source Development Lab to create software standards to meet the reliability and performance requirements for telecommunications applications.

AdvancedTCA and the PICMG

The Advanced Telecommunications Computer Architecture (AdvancedTCA) is a next-generation, advanced open architecture defined by the PICMG 3.0 standard. AdvancedTCA specifies interconnect technology, mechanical form factor, as well as power and platform manageability for telecommunications infrastructures like the media server and media gateway platforms, wireless network controllers, edge and access routers. It is scalable and versatile enough to take the industry through at least the next 10 years.

Intel worked extensively with all 106 PICMG members on developing and publishing the AdvancedTCA standard in only 15 months, and making it more robust than any standard in PICMG's history. The standard was approved at the end of December, 2002. The architecture is based on high-volume computing components with the reliability, high-speed interconnect technology, and other features necessary for telecommunications applications. With over 100 companies involved, there had to be numerous compromises, but the standard also maintained a great deal of flexibility to allow each vendor to innovate on the basic platform.

Initial TEM development efforts have shown that AdvancedTCA reduces development time and costs by 30 to 50 percent, and the benefits could increase as the ecosystem of AdvancedTCA-compliant component vendors grows. A number of industry leaders, including HP, Radisys, Motorola, Siemens, Lucent, NTT DoCoMo, Fujitsu and Siemens Computers have all committed to shipping products based on AdvancedTCA architecture.

Carrier Grade Linux and OSDL

The operating system choice for AdvancedTCA is Carrier Grade Linux. An advanced team of Intel system architects worked with Nokia to spearhead the development of a version of standard Linux with carrier-grade features starting in 2001. In 2002 we handed the specification over to a standards body called the Open Source Development Lab (OSDL), which evolved it to become the industry's Carrier Grade Linux specification. There is now an open source movement through OSDL's Carrier Grade Linux working group, and other members of Intel's telecommunications ecosystem are supplying carrier-grade Linux solutions on top of AdvancedTCA.

Many of the features of Carrier Grade Linux are aimed at improving system availability to meet the 24/7 or “five nines” (99.999 percent) availability requirements of telecommunications systems. The specification includes things like hardened drivers to ensure that a faulty driver won't bring the system down, robust file systems, debuggers and event-logging to help track system events that could lead to failures. There is also a real-time scheduler to handle soft real-time tasks performed by communications applications. The specification is evolving, and OSDL is currently working on version 3.0 of the Carrier Grade Linux specification.

Manageability Through IPMI and the SA Forum

Moving up in the system stack, communications platforms also require manageability. Intel is helping promote manageability in telecommunications platforms through two avenues.

First, we have joined with Dell, HP, and NEC Corporation to develop the Intelligent Platform Management Interface (IPMI), a specification designed to help platform manufacturers improve server manageability and lower operating costs for their customers. The specification defines common interfaces that allow IT managers to receive status alerts, send instructions to industry-standard servers and run diagnostics over a network versus locally at the server. Intel is moving these capabilities onto the board, to help lower the cost of delivery. In the AdvancedTCA specification, PICMG used IPMI as the cornerstone of the platform manageability.

Second, Intel works with the Service Availability (SA) Forum, a consortium of industry-leading communications and computing companies working together to develop and publish high availability and management software interface specifications. The SA Forum, originally called the High Availability Forum, was formed about four years ago by a group of telecom companies that wanted to create a high-availability solution with not just five nines, but six-nines availability. Intel is a founding member of the SA Forum, and is actively involved with the group.

The founders of the SA Forum formed multiple working groups to standardize various manageability interfaces. As part of this effort, Intel championed and co-developed a specification called the Hardware Platform Interface (HPI) with other companies, including Radisys and Motorola. This specification was later adopted as the standard platform interface for hardware management. You can think of HPI as an application programming interface for IPMI that allows high-level applications to access hardware information without the complexity of IPMI commands. The SA Forum has also developed an application-level interface standard called AIS. Intel has limited involvement because AIS defines manageability interfaces above the middleware level.

Recently, the SA Forum has also established liaisons with other standards bodies like PICMG, OSDL and the Distributed Management Task Force (DMTF) for greater collaboration and tighter linkages. For example, the SA Forum's HPI working group is mapping the AdvancedTCA platform to the HPI specification so that the full extent of AdvancedTCA's platform manageability can be exposed through HPI. Intel is leading this mapping effort. The mapping specification and the next revision of the HPI specification (version 2.0) are targeted to be ready around the June, 2005 timeframe. In addition, OSDL's Carrier Grade Linux supports an open-sourced HPI implementation called OpenHPI.

ICA: Building an Ecosystem

Industry standards are now playing a crucial role in making telecommunications platforms more interoperable and more affordable for TEMs and their customers. Innovations such as the standard mechanical-electrical form factor provided by AdvancedTCA, the next-generation modularity of PICMG's new Advanced Mezzanine Card form factor, manageability features from the SA Forum, and the operating system (OS) high availability features of Carrier Grade Linux are creating a standardized, modular foundation that allows communications equipment vendors to focus on the system-level and platform-level services they offer to their carriers.

The other critical element for equipment vendors is a complete and reliable supply of modular components based on these various standards. To meet that need, Intel has created the Intel Communications Alliance, a community of communications and embedded developers and solutions providers committed to the development of modular, standards-based solutions based on Intel technologies. To help suppliers develop standards-based components, Intel has started building its own AdvancedTCA-compliance reference deployment, building blocks, and compliance tests. Intel is presenting AdvancedTCA Interoperability Workshops (AIW) on how to use these building blocks and tests through the Intel Communications Alliance.

Intel has also created a Modular Communications Platforms (MCP) Design Guide in collaboration with industry partners like Motorola, Radisys and others. The MCP Design Guide has had significant industry impact, generating thousands of downloads from the Intel Web site. The Design Guide went through a major revision recently, and version 2.0 was released in November, 2004. Version 2.0 of the Design Guide can be downloaded from the Intel Web site.

Moving Toward Modular Platforms

Our biggest remaining challenge is still interoperability. As with any large-scale standards effort, the AdvancedTCA specification involved a lot of compromises in the first version. However, a great deal of knowledge has been gained through various AIWs and through the Design Guide, and the plan is to inject some of these learnings into the AdvancedTCA specification through various engineering change request/notification (ECR/ECN) releases. In 2004, Intel completed the first major ECN release for the AdvancedTCA specification, addressing several loose ends. The industry will keep improving the specifications with further releases over the next several years. AdvancedTCA version 2.0 is currently slated to be released in mid-2005.

The second challenge is to be sure other specifications, such as those from the SA Forum, are completed on time, because software manageability is a big challenge. Security, for example, is not completely defined.

The current version of AdvancedTCA is defined to provide headroom for at least two generations of platform technology, so we can produce the next two generations of platforms without breaking the current levels of interoperability and compatibility.

Summary

While there is still much to be done, the industry is already seeing the power of modular communications architecture. Modular components are already eliminating a great deal of duplication of effort across the industry. AdvancedTCA is lowering development costs and shortening time to market for telecommunications platforms, enabling TEMs to put much more value-add on the software side, which is enhancing reliability and availability for their customers.

Telecommunications providers are seeing the end results of standardization, getting more computing power for less money and lowering operating costs. For example, a single radio unit controller (RUC) setup used to take a whole rack, seven feet high, containing from three to 20 boxes of boards. It took a lot of power to run, and the complexity of configuring and managing all those boards added up to a lot of system management costs.

Today, two AdvancedTCA boxes can handle the same computing load. This means that a company like Lucent can manufacture two AdvancedTCA boxes capable of delivering coverage for one Manhattan exchange that used to take six to 12 boxes, and with higher availability and better service to end users. Clearly, modular communications standards are already a win for companies in the telecommunications industry and their customers.


More info

To learn more about some of the technologies and organizations mentioned in this article, click on these links:


About the authors:

Aniruddha Kundu is Principal Engineer with the Chief Technology Office at the Intel Communications Group. Kundu joined Intel in 1990 in the Intel Platforms Group. He has held various lead architect/design engineer positions. Beginning in 1993, Kundu worked as a lead platform architect on the first CMOS pipelined burst SRAM (PBSRAM) for L2 Cache and the Triton chipset, the PC SDRAM memory architecture specification and 100-MHz DIMM design guide, and the 440BX chipset. In 2000 he joined the Communications Platforms Group (CPG) and was a lead engineer for CPG's blade server architecture and many of the features for the AdvancedTCA specifications. Kundu has received eight patents on various computer system architecture areas, as well as two Intel achievement awards. He holds an M.S. degree in electrical engineering from Science College, India.

Udayan Mukherjee is a principal engineer and leads a team of architects for Intel's Modular Communications Platform Division, part of the Intel Communications Group. Mukherjee and his team drive Intel's technical effort in the Service Availability Forum and PICMG, as well as help define the architecture for modular platforms products based on these architectures. Prior to this role, he was the chief architect for Intel's Telecom Software Program, defining a Carrier Grade Linux solution for the telecommunications industry. Mukherjee was also involved in development work on blade-based highly available switched architecture for dense server and appliances. Before joining Intel in 2000, he spent 14 years at various companies, including more than seven years at Hewlett-Packard. Mukherjee has a bachelor's degree in mechanical engineering from India and two master's degrees in computer/systems science and industrial engineering from Louisiana State University.


Copyright © Intel Corporation 2005. All rights reserved. Reproduced by DeviceForge.com with permission. This article was originally published in Intel's [email protected] Magazine.


Other Related Stories:

 
This article was originally published on LinuxDevices.com and has been donated to the open source community by QuinStreet Inc. Please visit LinuxToday.com for up-to-date news and articles about Linux and open source.



Comments are closed.