For a decade, the telecom industry has been moving away from proprietary hardware and software toward commercial off-the-shelf (COTS) technology. 2005 will be a landmark year in this evolution, as ATCA systems reach a critical mass of support from hardware and software suppliers.

Advanced Telecom Computing Architecture (ATCA) defines a standard architecture that will extend COTS to the "bearer path" - the most demanding carrier networks - for switching and transport of voice and data. While ATCA mainly focuses on hardware interoperability between COTS suppliers, another set of emerging standards for managing this heterogeneous environment will be the catalyst to ATCA's rapid adoption. These management and availability standards are being defined by the Service Availability Forum (SAF) and are also maturing in 2005.

COTS hardware and software that is both ATCA- and SAF-compliant will enable network equipment providers and their service provider customers to quickly implement new high-bandwidth services to wireless and wireline end-users, such as multimedia-over-IP services running on 3G networks. These systems will be lower-cost, more reliable and easier to manage and maintain than the proprietary systems they replace.

Many pieces must come together to turn this vision into a reality, and 2005 is the year it is happening. A full ecosystem of standard components is now coming into place, including not only servers and switches but also carrier-grade operating systems, management and high-availability software, and application-specific I/O hardware.

A lower-cost infrastructure for 3G

ATCA is designed to let network equipment providers quickly and cost-effectively provide solutions using COTS carrier-grade technologies from leading component vendors.

ATCA, also known as PICMG 3.x, was developed by the PCI Industrial Computer Manufacturer's Group (PICMG) as a successor to the CompactPCI standard (PICMG 2.x). The advantages of ATCA over CompactPCI include far greater bandwidth (2.5 terabits per second vs. 3 gigabits per second), greater power and cooling capabilities, larger board area, redundancy, and integrated base-level system management.

The move toward COTS began with adjunct computing applications a decade ago, when NEPs began using COTS for handling functions such as network operations, administration, maintenance and provisioning. With the arrival of CompactPCI equipment, NEPs began using COTS for more critical parts of the infrastructure, such as call control and new calling services like caller-ID and push-to-talk.

However, this second generation of COTS technology lacked the bandwidth and scalability to handle the most critical part of the infrastructure, the voice and data traffic itself. ATCA delivers those capabilities. It enables cost-effective deployment of 3G wireless networks that can process massive amounts of compute-intensive data for applications such as multimedia streaming and routing.

To base their strategies around ATCA technology, NEPs need not just individual components but an entire ecosystem of vendors delivering in each key area of technology. Vendors need to work together to make sure that all important pieces of the technology are available and interoperable. The good news is that this is happening today. Key components that will hit the market this year include:

• ATCA-compliant servers.
• Increasingly sophisticated and reliable off-the-shelf carrier-grade operating systems. Solaris 10 is now shipping, and a new version of carrier-grade Linux is expected in 2005.
• SAF-compliant network management cards and high-availability software.
• Call processing, service creation and service enhancement software. While software in use on CompactPCI systems should run on ATCA systems without modification, the software needs to be stress-tested to ensure it can handle the massive additional throughput capacity of ATCA. Software vendors are engaged in this process now.
• Application-specific I/O hardware, including network processors, digital signal processors and gateways.

Stay focused on this space. A wide range of hardware and software providers will be delivering ATCA solutions this year in the areas of protocol stacks, digital signal processors, application software and more.

Beyond ATCA: Systems management, high-availability middleware and systems integration

Though ATCA is a considerable advance over CompactPCI, it still leaves a lot of work to be done, either by the NEPs or their system suppliers. ATCA specifies only base-level system management, leaving it up to NEPs or their suppliers to build higher-level management and reporting tools. In addition, turning a set of ATCA-compatible components into a high-value solution will require considerable systems integration work.

Prior to CompactPCI, NEPs such as Nortel and Siemens had to build everything from scratch, from the racks to the network hardware to the system management software. CompactPCI delivered a basic level of standardization, specifying standards for chassis, blades and interconnect.

Although with CompactPCI a user could plug a standard I/O blade into a standard rack, there was no standard way for the blade to report a problem to the system administrators. Under CompactPCI, NEPs were left with the task of creating a management infrastructure for disparate hardware components.

ATCA has base-level system management built in for both hardware and software. It defines the physical form factor, the data connections between the boards, and the method for powering the boards. But there are no higher-level capabilities for addressing system management on the rack level or the solution level. ATCA doesn't provide standards to enable a multi-vendor network of these systems to interoperate in isolating faults, taking recovery actions, distributing software, managing inventory or conducting other maintenance activities.

A partial solution is under development from the Service Availability Forum. SAF is developing a set of standards to address management of heterogeneous systems, in a manner that is complementary to ATCA. The SAF standards will allow NEPs to model an ATCA system using components such as blades, switches and chassis from multiple vendors, in a way that allows management applications to manage the system in a vendor-independent manner.

SAF is further complementing the ATCA ecosystem by defining standard interfaces for high-availability middleware. The combination of standards-based middleware with ATCA provide a fully-integrated COTS solution to NEPs, allowing them to reduce their investment in proprietary middleware and focus on core added value in application software and services. SAF is quickly gaining maturity in the middleware space, as its standards get rolled out and adopted by the industry.

There also remains a need for systems integration. Think of ATCA as a standard for manufacturing bicycle parts. A person can build a bicycle if someone hands him the bearings and hubs and spokes and gears, because all the components are designed to work together. But he could build the bicycle much faster if he could buy pre-built wheels, a pre-assembled gear cartridge, and other modular components.

Technology suppliers and systems integrators will step into this role, delivering pre-kitted, modular ATCA solutions that relieve NEPs of much of the systems integration tasks. This will dramatically speed time-to-market and reduce development costs for NEPs.

Raju Penumatcha is Vice President, Netra Systems and Networking (NSN), for Sun Microsystems Inc.