Siemens AG selects Teradata to help it on its way to the 'Internet of Trains'

Siemens AG picked Teradata over HP Vertica, IBM and Oracle because it was 'the first to understand the need for an ecosystem'

Siemens AG, the German engineering company, is using Teradata's data warehouse technology because it was "the only system on the market that solved all of Siemens' problems", according to Gerhard Kress, director of mobility data services at Siemens AG.

Kress was speaking to Computing at Teradata Universe in Amsterdam last week, and he explained the need to have a data warehouse and analytics solution to enable Siemens to gain insights into the development of its trains, in a bid to eventually create "the Internet of Trains".

He said that the company was looking for a system that could scale in the petabyte range in order to achieve this.

"That means it allows for data discovery, so analytics where you don't use the standard approach - it's not a business intelligence [method] and that means you have to be able to push analytics into the database," he said.

"Three years ago when we did a market sweep - Oracle, IBM, HP Vertica, and of course Hadoop on its own - not all of them [could do that] and we selected Teradata."

Kress explained that the key factor in the decision was Teradata's Unified Data Architecture (UDA). He said that having the three components of a data warehouse, Hadoop and Teradata's data analytics platform Aster was "fantastic".

"I believe that Teradata was the first one to see that you can't just have one system that solves all of your problems. You have to be a part of an ecosystem. And if you compare the division in the UDA between Hadoop and the data warehouse now to two years ago, it has already changed and it will continue to do so because both are evolving, and we're using both extensively," Kress said.

Internet of Trains

Kress said Siemens has been applying Internet-of-Things-type technologies to trains long before rival GE, and he took a swipe at the US firm.

"Siemens has a bit of a challenge in that we market things we already have in operation, whereas in the US, they market it and then they build it," he said.

He went on to explain that the firm has a platform that integrates devices from the field, and that is used in healthcare, trains and wind turbines. In total, the company has about 300,000 connected devices.

However, it needed to rebuild its underlying platform because the existing system had outgrown what Siemens originally believed it would need to do.

"It grew too fast and we needed to upgrade the platform a bit more. At the moment the IoT is an important element for us but we gave it a slightly different twist - we call it a web of systems," he said.

He explained that this is because the devices that send out the data have no logic.

"In IoT you may have chips and they will communicate, but in our case it could be that a train has a large computing capability, and that's a lot of processing that's required," he said.

He added that the volumes of data involved could never be transferred over the air to the wayside, and so all of the processing is done on the train itself.

"There is a lot of intelligence on the train system itself, that's why we call it a ‘system' and not just a ‘thing'. In future, trains will start talking to other trains about conditions, and things like ‘what is the most energy efficient way of driving'," he suggested.

But this will require a huge step change in the train industry, which Kress said is very conservative owing to the long timescales involved.

"If we sell a train, we envisage that it will be operational in 30 to 40 years, and in some cases longer. Today, we are signing a contract in Germany where we sell trains together with 32 years of maintenance at a fixed rate. The longest-running service contract we have is a 50-year-old contract. So these are heavy investments and it creates a problem because a lot of stuff out there is old and it is still operational. It was built 20 years ago when the technology was very different," he said.

Upgrading or changing software can also be problematic, because the company has to go through certification with authorities, which can sometimes take up to a year.

"That's why we start building our platform on the wayside, so not on the trains because that way we can do what we want, we don't need to bother with certifications and I can just change the algorithms and iterate a lot faster," he said.

Kress said that the overall platform could allow Siemens to perform analytics on its data, enabling it to reduce the cost of rolling stock fleet maintenance, minimise train downtime and improve fleet availability.