The xTCA eco-system – what is your limit? By Sven Freudenfeld, Business Development Telecom Kontron The current market condition has shown that using COTS components and platforms, in particular xTCA, to create key network elements are an even more valuable approach than in-house proprietary designs. Since its creation as a standardized hardware platform in 2005, more xTCA components have become commercially available in the market, providing the “Freedom of Choice” for vendor selection. In theory it is the right approach to get the best designed components from the best vendor at the best price in a competitive eco-system. However, some consideration and homework still needs to be done to ensure there are no long-term impacts in the platform lifecycle. The fact that these components are standard based means there are a wide range of choices. Conversely, having many choices are what can also be difficult to make a correct decision. In fact, since it is standard based, any company in the market can start developing and provide components and platforms for NEPS from a small design shop to a top-tier embedded computing provider. Since the standardization process is completed, early adopters in the market launched and released xTCA based platforms and they have been shipping products for the telecom network since 2006. The first round of xTCA-based product rollouts were not entirely taken advantage of to be multi-vendor platforms. They had been become single-sourced xTCA platforms. With consolidations and economic recessions in the market, some smaller, not so stable xTCA vendors have disappeared, while others went through acquisitions and, as a result, a number of components became EoL for the early adopters. Not a desirable scenario from the NEP’s perspective. With field-deployed systems, it is not always a simple upgrade to move from one vendor to another vendor. Both NEPs and carriers need to avoid fork-lift upgrade operations in order to maintain network functionality and backward compatibility, which is more crucial whenever it comes down to choosing the right vendor. The recent Light Reading Communications Ecosystems Conference (CEC) event that focused on ATCA, http://www.lightreading.com/live/event_information.asp?event_id=28913&survey_id=521,revealed the latest developments within the standards organizations, and gave an update on the strategies of the key players in the industry. Noticeable too is that with the newly announced Intel Xeon 5500 processor series for the embedded market, a variety of compute blades are hitting the streets. However, with the increase of memory bandwidth, the variations of blades are within a 200W power envelope to a slightly elevated 300W and beyond per blade. The key questions are: where in the network will these 300W and beyond standard-based blades find their homes? What CPU and memory configuration will take full advantage of the new architecture? In other words, what is your limit? One important thing in using these newly released Intel Xeon 5500 blades is the backward compatibility and the insurance that these blades can also replace “legacy” CPU blades used in a currently deployed 200W capable chassis with full NEBS compliance. The overall ATCA system lifecycle should not be impacted when selecting newly available CPU blades on the market. As a second consideration, interoperability between chassis and blades has become even more crucial when looking at these ranges of variations. And we haven’t even looked at the AMC usage model for ATCA yet. The Communication Platform Trade Association (CP-TA), http://www.cp-ta.org ,promotes the xTCA ecosystem and has created baseline tools and interoperability documentation to provide mainstream ATCA adoption for standard-based blades. However, the challenge remains when integrating blades in a chassis that are exceeding the power envelope in a multi-vendor environment. One news announcement at the CEC event is the creation of a subset of the xTCA standard as an extension to address the non-NEBS and ‘beyond the CO’ environments. Being proposed is a power budget beyond 600W in a double-wide ATCA CPU blade approach with an extremely high compute density and a low cost point. Probably, this is technically feasible but this will require an ATCA chassis that will not be used in a CO environment but would be dedicated to “ATCA in the Datacenter”. This demonstrates that more work is to come for CP-TA to address interoperability in those markets. Besides that, I just returned from the MicroTCA Summit and this year the main interest was clearly the use of xTCA outside telecom and, in particular, government and military applications. It is great for the ecosystem to see how ATCA and MicroTCA finds its market in these areas of applications, and the work within PICMG to address the rugged environment will help even more to push xTCA mainstream. xTCA is the ideal HW architecture for many applications in - and outside of telecom. When using it, what is your limit?