The big optical switch

It seemed to make sense to start by asking some fundamental questions about the structure of our network. Which network services would we need in future? How did we intend to provide them? What system requirements could we derive from this? At the same time, we got together with manufacturers to take stock of the latest technology and look ahead to what was coming. Developments in optical communication over the past ten years have been truly remarkable, and the potential for innovation is far from exhausted. The following new features and functions are worth mentioning:

• Miniaturisation: transponders working at 10 Gbps (gigabits per second) have shrunk from the size of a book to the size of a matchbox, and their power consumption has been cut by more than a factor of ten.
• Optical signal density: whereas our old system supported 16 different colours, the state of the art is now 80 parallel optical signals (wavelengths or "lambdas"). In addition to vast capacity reserves, this brings lots of advantages from an operational point of view.
• Photonic switches use tiny mirrors to steer each individual lambda to a particular outlet without the need to convert it into an electrical signal.
• Broader channel bandwidths: with the simple on/off switching of the light source replaced by a more complex modulation scheme, transfer rates of up to 100 Gbps per lambda are possible.

Flexibility: lasers can now be set across 80 different lambdas with a simple configuration command. This, combined with photonic switches and the new generation of tunable filters, means that cables no longer have to be repatched locally in many cases.

How we approach the individual phases of the ALPSTEIN project:

2012: gathering information on the state of the art

Manufacturers of optical transmission systems tell us how far the technology has come and where it is headed. They present their proposed solutions for a configuration we have defined and estimate the costs.

September 2012: concept, cost plan, proposal

The "100GLAN Concept Plan", a joint effort by the Global LAN team, is completed. We outline what the network will look like and estimate the construction and operating costs. We draw up a proposal to put forward to the Foundation Council.

November 2012: approval

The Foundation Council approves our project plan and sets aside the requisite funding. We write a detailed set of specifications as a basis for hardware procurement and renewing the fibre-optic infrastructure.

April 2013: project organisation

We assemble a project team comprising people with a wide range of skill sets. We put a great deal of effort into the tender documentation to speed up the subsequent evaluation of tender offers and to ensure that the decision-making process is transparent.

 June 2013: publication

The tender is published. Tendering firms are invited to submit questions, which SWITCH answers promptly, forwarding its response to everyone participating in the tender.

August 2013: shortlisting

SWITCH has studied the 12 tenders offers it received. The project team locks itself away for three days until it has drawn up a shortlist. Those who have made it onto the list are invited to present their solution to SWITCH.

October 2013: contract signing

The decision is made, and the contract with ECI is signed. Implementation begins. The 1,000-km fibre-optic ring between Geneva, Zurich and Lugano is to be migrated first.

November 2013 – April 2014: preparatory work

With the detailed design for the first construction phase in place and the necessary materials ordered, the requirements for the migration must be met. New rack space has to be organised at all locations, and fibre-optic infrastructure has to be converted to allow two-fibre operation – all of this without zero downtime. At the same time, training is provided for the new system, and the manufacturer conducts lab tests.

May 2014: start of installation

The first stretch we tackle is Zurich-Lugano. The installation engineers set up all the nodes along the route and configure them so that they are visible in the network management system. The manufacturer's technicians then take over and configure the network from their offices. The final, critical step is going live, which has to be done "on the fly" due to the lack of new fibre-optic lines. All clients must remain contactable at all times, only the redundancy can be done without. We form three teams in each case, each including a technician from the manufacturer to calibrate the optical system and a SWITCH employee to look after the local switching. This way, two to three sections of the route can be switched over every day.

September 2014: end of the first rollout phase

Much of the required functionality is not verified until the testing phase. However, confidence in the new system is already very high because it has been operating reliably for months.

Mid-2015: full operation

The system is scheduled to be in full operation by the middle of 2015. The new backbone is so flexible that we can add extra capacity at very short notice. New client locations are connected via the most cost-effective route to the nearest backbone site, if possible using existing optical fibres. Sophisticated passive optical filters have been developed that allow us to use fibres more than once, i.e. for the backbone system and local connections at the same time. We can only guess what demands will be placed on the network in ten to 15 years' time, but we have a sound basis on which to build efficient solutions in the future.

The birth of an optical node

This series of photos documents the work carried out for the Alpstein project at the University of Bern on 5 June.

This article appeared in the SWITCH Journal October 2014