By: Dr. Markus Leberecht, Force Computers
Great are the cost and resource savings to be gained by using the same high-availability system architecture across multiple applications. This common platform approach is a focal point in ATCA.
The winds of change continue to build momentum in the massive communications infrastructure market. The convergence of data and communication networks is ramping up, and next-generation wireless networks are starting to deploy. At the same time, bandwidth requirements for IP services are steadily increasing. These trends are tasking both Telecom Equipment Manufacturers (TEMs) and Network Equipment Providers (NEPs) to craft new types of equipment at ever-increasing densities and at lower cost. Such challenges are driving both Telco operators and TEMs/NEPs to focus mainly on their particular differentiating factors. Outsourcing provides a viable means to achieve such goals within budget parameters.
All that has spurred interest in the concept of employing one single, outsourced platform technology and outsourcing arrangement as the common denominator across many product lines. The Common Platform concept implements a single high-availability system architecture across all hardware and software layers. This enables typical equipment manufacturers to efficiently execute a wide variety of communication applications and network elements. In the access and edge networks of the wireline and wireless infrastructure, those elements can include either server-type or control and data plane architectures.
Using the common platform approach results in cost savings at multiple levels ranging from more efficient design, harmonized logistics and less training to taking advantage of volume usage and high-volume effects on unit pricing that are not possible with point solutions.
Common Platform Enables Reuse
Using a common platform aids the whole value chain. First, it simplifies logistics. Basing the system on a single, versatile platform technology, a common platform makes frequent reuse of the same types of components. Meanwhile, a well-designed common platform helps leverage a significant amount of pre-integration work. Various features can be spun-off from one common platform application into another.
A common platform strategy also reduces the depth of the supply chain through the purchasing of pre-integrated platforms rather than components on various levels. It frees up resources to be able to handle the greater flexibility and choice of a standardized base—instead of building everything from components on your own, choose from multiple suppliers of pre-integrated platforms.
AdvancedTCA Common Platforms
A unique aspect of the Advanced Telecommunication Compute Architecture (ATCA), as defined by PICMG, is that it specifies the idea of a common platform. ATCA is primarily a packet-based, switched blade architecture that defines a fully redundant system including shelf management and remote access to each system component via Intelligent Platform Management Interface (IPMI). The data transport technology utilizes 2N-redundant switches, such as the Dual-Star Base Interface, the Dual-Star Fabric Interface or other Fabric options for future use. The node boards or blades are 8U in height, with 14 slots able to fit in a 19-inch rack.
In ATCA, node boards are hot-swappable with the open standard securing interoperability of boards and systems across different vendors. Power input is -48 VDC adapted to the typical central office environment. On a technology level, it is ATCA’s goal to support the migration toward converged networks by offering an HA-proof modular computing architecture. Advantages quoted for ATCA include technology headroom, engineering reuse, volume economics, component interoperability and application flexibility.
With that in mind, AdvancedTCA can reduce the expected total-cost-of-ownership (TCO) at all levels, even up to the operator level. That assessment is supported by independent market studies, including one by Yankee Group that estimates between a 20 and 35% reduction in capital and operational expenditures for TEMs and operators.
In order to specify common platform requirements, it is necessary to clearly separate application-specific platform features from joint qualities. Two generic application scenarios comprise the overall Common Platform direction: a server-centric scenario mainly targeted toward control and management plane applications, and one oriented toward data plane usage. A common platform should combine the shared characteristics for both scenarios while ultimately supporting a mix-and-match combination with additional components for multi-service applications. Figure 1 shows Force Computers’ ATCA Common Platform Concept, which supports both server-centric and data plane applications.
Both server and data plane application scenarios share a central-office environment. As a result, they require such capabilities as redundant networking infrastructure on the backplane within a highly available system. Such systems must support the typical “five nines” requirement of telecom installations. ATCA systems furthermore provide redundant out-of-band shelf management and can host a variety of line-interface as well as processing blades.
Software technologies working adjacent to ATCA also form a vital part of the common platform approach. These are represented by the Open Source Development Lab’s (OSDL) Carrier Grade Linux (CG) for blade operating systems, and the Service Availability Forum’s (SAF) middleware interface specifications such as AIS and HPI. With these components, a common platform consists of the underlying ATCA shelf hardware, at least two general-purpose processing blades running CG Linux with an SA Forum-compliant middleware on top for redundant HA management. Such functionality monitors the availability of system resources and eventually controls failure isolation and recovery mechanisms.
Additional board content differentiating this base platform from one or the other application scenario occupies the remaining slots. Typical server-type telecommunication applications that can be built with such base platforms include a Softswitch, a Home and Visitor Location Register (HLR/VLR)—used to facilitate identification and authentication of roaming mobile users in wireless networks—or a Media Server.
A Softswitch represents the call-control part of a converged network within a Media Gateway or Media Gateway Controller. It is tied mainly to signaling communication protocols, such as H.323, SIP, MeGaCo, SIGTRAN or SS7. The core component of the Softswitch’s software architecture will be hosted on general-purpose processor blades like those that also host the HA manager. By interfacing to the HA manager, this software can be made highly available as well. In a pure IP environment, the ATCA base and fabric interface networking may directly connect the system into the WAN, while with OC-x link technologies and specialized line interface cards are incorporated into the system.
A Home Location Register and its counterpart, the Visitor Location Register, represent geographically distributed subscriber databases for identification and authentication of roaming mobile users in wireless networks. Again, communication toward HLR and VLR is of the signaling type and requires similar blade content as above. For larger scale data storage, i.e., beyond the 300 Gbyte mark, these systems typically feature access to a central, redundant storage subsystem. ATCA offers Fibre Channel (FC) technology over the backplane for this purpose, and FC-capable processor blades can utilize this to interface to FC storage modules, also hosted inside the ATCA system’s blade slots.
Media Servers, meanwhile, are used mainly for audio and video content retrieval; these systems typically forward their payload in IP packet form to appropriate Media Gateways. The two main requirements for this application type are media-oriented processing power, which today can be found also in general-purpose processor architectures, and a sufficient media repository. Similar to the application above, a media server implementation can host FC-based storage controller blades as well as FC storage blades.
As explained earlier, data-plane application scenarios differ significantly. ATCA’s target area lies in the wireless networks’ access and mobile core domain (RNC, SGSN, GGSN, Media Gateway), as well as in the wireline access domain (DSLAM, BRAS, Router). ATCA responds to the need for physically separated control and data plane communication paths in these scenarios by providing the fabric interface as a second, dedicated high-speed network with flexible topologies. Exemplifying the applications in need of such a scenario are a Radio Network Controller (RNC), a Signaling GPRS Support Node (SGSN) or a Media Gateway.
A Radio Network Controller represents the interface between the UMTS Radio Access Network and the circuit-switching and packet UMTS Core Network, and controls NodeBs, the UMTS base stations. The resulting interfaces are thus of various types, including multiple DS-1 links or alternatively channelized OC-3 interfaces for NodeB communication, OC-3 to OC-12 interfaces to both the 3GMSC for voice traffic and SGSN for packet traffic. On the data plane, such a system performs access concentration across a large number of communication protocols. Next to the dedicated system hardware for the additional Fabric interface, line interface cards of different types and network processors (NPU) will be utilized for protocol processing.
In contrast, a Signaling GPRS Support Node (SGSN) provides routing of wireless packet data into an operator’s IP network from a number of RNCs. Next to the RNC-opposing interfaces of OC-x type, the SGSN needs to support signaling interfaces—like SS7 on DS-1—and IP data traffic for which Gigabit Ethernet is typically used. For the latter, ATCA’s generally Ethernet-based backplane communication (Base Interface is Ethernet by default) comes in handy and can be forwarded directly from the ATCA switch uplinks when the operator’s IP environment can be accommodated through network address translation.
Basically the data plane counterpart to a Softswitch, a Media Gateway provides the capability to transcode TDM voice circuits into voice-over-packet protocols such as VoIP. As it is controlled from a Media Gateway Controller, which may be located outside of the MGW’s system premises, it is a pure data plane architecture and interfaces to both the packet and circuit-switched worlds with either line interface cards or the switch uplinks. DSP processing resources on either dedicated blades or line card blades are responsible for the encoding and decoding of payload data.
Decomposing the Common Platform
As described earlier, a suitably designed AdvancedTCA platform with the necessary amount of pre-integrated platform software can serve as the foundation of a common platform program. Figure 2 shows a sketch of the hardware architecture of such a system by Force Computers. The base platform contains a number of minimally required components such as redundant shelf management controllers, each as separate field replaceable units, designed to provide remote system management by SNMP, plus platform monitoring and control through the SAF HPI interface. Redundant base interface switches in the system provide redundant 1 Gbit/s per slot, Nx 1 Gbit/s as redundant uplinks. The system’s Intel Pentium4M-based node blades support PMC mezzanines for easy configuration of additional line interfaces. Alignment to a number of industry standards with convergent, compatible and complementary vision gives the pre-integrated base platform its strength. Figure 3 illustrates those open standards-based software layers.
With a carefully chosen common HA base architecture based on an open, modular and reusable ecosystem, equipment manufacturers and carriers today have the opportunity to concentrate on and amplify their differentiating business and technology factors. AdvancedTCA together with a number of like-minded accompanying industry specs enables this common platform approach across a wide variety of central-office applications. Companies supplying ATCA products can, in turn, offer differentiated outsourcing and integration levels adapted to the actual business needs of their TEM and NEP customers.
© 2009 RTC Group, Inc., 905 Calle Amanecer, Suite 250, San Clemente, CA 92673