VME – Live or Let Die?
The Embedded Tech Trends (ETT) conference, organized by the VITA standards organization and held last month in New Orleans – a city which folklore says has the reputation for dark powers, Voodoo rituals, and magic. It was also a filming location for the James Bond movie “Live and Let Die” and one of the fun parts of these Bond movies is the new gadgets given to Bond to help his fight against evil. So, New Orleans seemed an appropriate location for the some of the world’s leading providers to the embedded markets to mingle and promote their latest magical innovative technologies for tomorrow’s applications.
When we stop and consider the rate of change of technology, we can quite rightly be amazed to realize how far we have come in relatively little time. We talk of the latest and greatest backplane communication and networking speeds that can be achieved, with and without the wizardry of software and firmware products that are available. No longer the restriction of one operating system (OS) on single-core processors; today, most of the processors we use are multi-core and capable of supporting multiple OSs by using bare metal or hosted virtualization products.
Additionally, modern OSs can schedule applications to be automatically spread across the different cores to improve loading and reduce performance bottlenecks.
And it is not just the boards that have improved to deal with this ever-increasing desire and need to collect and process data.
The latest serial fabric based architectures support increasingly higher bandwidths across the backplane to enable the collection, processing, and then transfer of ever increasing amounts of data between boards and systems. VPX, for example, supports backplane architectures today that can provide transfers of as fast as 10 GBaud per lane for use with 10 Gigabit Ethernet and gen 3 PCI Express fabrics. This has driven connector technology capable of supporting both the higher data rates and the extra level of ruggedization designed in as standard with specifications like VPX.
Performance is greatly improved, but the complexity and price of the system increases and so this is not always the best or most appropriate solution.
In our embedded market sectors, we are always keen to talk about the latest and greatest and foresee what will be available in the future. We need to keep pushing the boundaries in order to advance. But, it is perhaps important to remember that the older architectures are still at the heart of many of the applications in production today within the embedded markets because they typically have very long life cycles: the Boeing B-52 Stratofortress first entered active service in 1955, with many still flying today and after a recent upgrade, are not expected to be retired until 2040.
Whilst this might be an extreme case, almost all embedded applications require long life-cycle products and need to be supported for many years. So it’s not surprising that older architectures continue to be needed and used for many years after the introduction of the newer technologies.
Take VME for example. This bus was “invented” in the 1970s and first ratified as a standard in 1982. Every year, it seems there is a prediction that it will be replaced by newer architectures. In some embedded markets this has been the case: for example, VME was once widely used in medical and industrial applications.
However, the defense market requires much longer life-cycles and has proved to be extremely reluctant to replace VME deployments. In fact, VME evolved to provide improved but upwardly compatible formats and continues to be a significant feature in today’s embedded applications when newer, better, and faster technologies are available. In 2016, IHS Markit teamed up with the VITA standards organization to deliver a market report which concluded that the total market for VME boards and systems is still larger than that for VPX boards and systems. IHS Markit is now forecasting the crossover when VPX shipments outperform those for VME to be 2019. More detail on this survey can be found at. Time will tell how accurate it is, some previous surveys had predicted a much earlier and faster decline for VME shipments.
We must also keep in mind that our customers have budgets too, and it is not always viable to throw out their investment in a VME-based system. If a system can be maintained and enhanced and does not need all the new capabilities that are available with newer architectures, then a “simple” upgrade will improve the company’s return on their product development costs.
Take for example a company supplying equipment for use in surface ships and submarines. This company has probably taken many years to develop and certify their systems based on VME and has made a significant investment into that product. Much of the I/O will remain the same to interface between other systems installed on the vessels – but some new requirements will be included within technical upgrades such as support for newer storage devices and improved processing performance. It might seem like an opportunity to introduce VPX, but there are some very good reasons to remain with VME, including:
- The cost and time of developing and certifying a completely new system.
- Compatibility and availability of all the parts of the system including legacy I/O.
- Viability of a forklift chassis upgrade. The final customer may not be willing or able to refit a vessel with new cabling, backplanes and chassis needed for a new architecture. This could be due to the amount of time it would take and/or the expense it would entail. Government budget restraints have in the past meant that upgrades rather than replacements were the order of the day.
Companies like Concurrent Technologies continue to provide new products based on the VME architecture. New features are important of course, but it is equally important to offer plug and play replacements especially where specific I/O interfaces are concerned, and to minimize the changes needed at a firmware and software level.
As embedded board vendors, we will continue to strive to offer the latest products as soon as possible. However, parts of the embedded market need technologies that will continue to be manufactured and supported for many years and this is very different from the ‘throw away’ commercial gadget market.
For sure, we need the new Bond gadgets with the wizardry and magic of new technologies. But unlike the Bond film title, customers with long life-cycle applications need the old architectures to Live and Let Live.
Security issues….Since James Bond rarely managed to return any of his secret gadgets intact, let’s hope those inventors were using some of the security options that we build into our current range of embedded products regardless of architecture, and that his gadgets really did self-destruct before falling into the hands of his enemies!