The times they are a-changin' ... still

When Bob Dylan wrote that song 40 years ago he was likely thinking about the dramatic social and political changes that were happening in the 1960s, but a similar change in the technology of big networks is happening now – the way they are built, used, and the business models that support them are all being transformed. Terms like Network Functions (NFV) and Software-Defined Networking (SDN) are the latest additions to the alphabet soup of terms used by traditional telecom service providers and in the datacenter. The architectures used by both of these industries are beginning to merge into one where services, storage, and computing are provided to customers that want to sell something via e-commerce or share cat videos or corporate spreadsheets. With the increasing use of mobile devices, programs and data will not be on your desktop, but “in ,” and accessible from anywhere.

This column was going to focus on introducing some of the concepts involved with SDN and NFV, but my good colleague Curt Schwaderer did a far better job than I ever could have in his excellent article on page 12, “ () explained.” I highly recommend you read it, and also the European Telecommunications Standards Institute (ETSI) white paper he references.

One basic value proposition of this new approach to providing services is that as many functions of a traditional network as possible are virtualized so that they can run on common hardware, meaning lots of servers that are physically similar and can be purchased, deployed, and maintained en masse. This is somewhat different than today’s approach, where lots of different and specialized boxes from a variety of suppliers must be made to work together to provide end-to-end services.

Does this new world bode the end for specialized platforms like () because they can be replaced with racks and racks of cheap, 1U “pizza boxes?” The answer is no, because platforms like ATCA offer important features that Tier1 providers demand – High Availability (HA), extended temperature operation, Network Equipment-Building System (NEBS) compliance, sophisticated platform management, and support for a wide range of high-speed copper and optical interconnects.

That is not to say ATCA doesn’t need to evolve to provide increased capability; it does. Due to the inexhaustible appetite for more processing power and more storage, platforms will need more real estate for processors, memory, and storage. Simply put, the industry will need bigger boards to provide that real estate.

The ATCA community has realized this, and is nearing completion of a fairly major enhancement to the core ATCA standard. This new specification is known as “ 3.7” or “ATCA Extensions.” It expands the packaging definitions to include dual-sided shelves, where ATCA boards can plug into either the front or back of a double-deep rack and interconnect through the backplane. In addition to this, the Extensions spec also allows for something called Extended Transition Modules (ETMs) that are essentially a front board-sized circuit board that connects to a front board via Zone 3, much like a standard Rear Transition Module (RTM). There are many variations of interconnects allowed, but Figure 1 gives a general idea of the concept.

21
Figure 1: Pictured here are representations of the front board, Extended Transition Module (ETM), and dual-sided backplane outlined in the PICMG 3.7 specification, which is nearing completion.

Importantly, PICMG 3.7 provides a much more detailed definition of, and support for, double-wide modules than the original specification. These can support multiple processors, bigger heatsinks, cheaper full height memory modules, and multiple disk drives on a single assembly if desired. PICMG 3.7 also defines requirements for typical datacenter environments in addition to the telco central office. Double-wide modules can support up to 800 W of power dissipation if the shelf is built for that. AC, as well as traditional -48 V DC, power environments are also supported.

All of this additional function means that the platform management infrastructure needs to be expanded and enhanced, and Mark Overgaard of Pigeon Point Systems (PPS) provides detail as to how PICMG 3.7 systems can be configured and managed in “Upcoming AdvancedTCA Base Extensions specification benefits current and future systems – including for services,” which begins on page 14.

Other things are a-changin’ also. This issue marks the last one named and Systems. Starting with the next issue, which will debut at the Conference in , Germany in February, this publication will be re-named and re-purposed to cover the entire range of PICMG platforms, both currently in use and on the drawing board. The new book, PICMG Systems & Technology will be more inclusive in scope and reach. And lest you think that PICMG 3.7 is the end of our planned work for building bigger and more capable platforms, the first issue of the new publication will provide a peek at the roadmap for the next large PICMG platform, GEN4, which will provide more capability and horsepower than exists in any open platform on the market today. I’ll leave it at that for now.