Where are data centre technologies coming from and where are they going?

The rush for new technologies to meet bandwidth demands in data centres is clear, but how and why is it happening? asks our regular contributor Willy Rietveld, from TE Connectivity (W.Rietveld@TE.com).

  • Monday, 27th April 2015 Posted 9 years ago in by Phil Alsop

TE Connectivity is active in all areas of the network. In theory, you could send an e-mail from your laptop to somebody on the other side of the world and all network parts used could come from the company – including the undersea network. As I will go on to explain in this blog, this gives us a unique insight into how new data centre products are being designed.

Unsurprisingly, for subsea communications we try to get as much data (bandwidth) through each SM (single-mode) fibre, over the longest length possible, because of the high costs associated with deploying cables on the ocean floor. Our SubCom business designs, manufactures, installs, and maintains transoceanic-length undersea fibre optic cable systems that have the capability of transmitting 10,000 gigabits per second over each fibre pair. Since a typical cable system might have four fibre pairs, the capacity of each individual cable is ~40 terabits per second and yes, that’s a lot of capacity!

On the other side of the spectrum, in our consumer division we continue to see innovation accelerating at an ever-increasing pace. The latest goods to hit the market are wearable wireless technology products or ‘wearables’. Wearable computing applications include everything from fitness trackers and health monitors, to smart watches. Some interact with your smart phone, while others add a brand new functionality, such as augmented reality. What’s interesting is that these use cases can already be seen in data centres. For example, installers can view installation instructions hands-free, while data centre owners can guide maintenance people from the comfort of their desk.

Somewhere in-between we have the HPC (High Performance Computing) world. Here, we have seen a migration from MM (multi-mode) to SM, and a shift from 10Gb/s transceivers to 25Gb/s transceivers. Again, this also appears to be happening in today’s data centres. In the HPC world, the next steps are 50Gb/s integrated photonics and waveguides. How long will it be before we start seeing this in data centres as well?