The network may have its Achilles heels, but SWITCH knows where the vulnerabilities are and how to protect them.
It runs and runs and runs, 24 hours a day, 365 days a year.However, it is very much the case that network components fail from time to time.SWITCH has taken precautions to ensure that its clients never notice when this happens.Let us take a look at where the network’s weak points are and how SWITCH deals with them.
Fibre-optic cables are very reliable, and experience has shown that outages are extremely rare.Nevertheless, cables are occasionally damaged by building work, accidents or rodents.Depending on the nature of the damage, it can take days or even weeks to rectify a fault.
SWITCH plans redundant connections to the SWITCHlan network wherever possible.We try to find two geographically separate routes for the fibre-optic cables with a view to preventing lengthy outages for our clients.
Even the best hardware can sometimes fail.SWITCH employs high-quality hardware wherever possible,but components such as power supplies, line cards, transceivers and optical amplifiers can still malfunction once in a while.
We offer our clients SWITCHlan connections with all electrical components duplicated.If one router fails, for instance, the second automatically handles all data traffic.That way, clients are never offline.
Most outages are caused by power cuts,despite the fact that Switzerland’s electricity supply is very stable.
Wherever possible, SWITCH runs all active components with two independent power supplies.The typical configuration is one mains power supply and one uninterrupted power supply or UPS.We also try to separate the two routers from each other geographically for redundant SWITCHlan connections.This also tends to reduce the dependency on a reliable power supply.
In addition to SWITCHlan connections, connections to the international research networks and the commercial Internet are also at risk from hardware defects, fibre-optic outages and power cuts.Problems may also arise with routing, such as black holes or configuration errors.
SWITCH now has a redundant connection to international research networks as well as four uplinks to the commercial Internet in Zurich, Geneva and Basel.There are also countless peering connections in Zurich, Geneva, Basel and Amsterdam.This large number of uplinks improves quality by shortening pathways.If one or more peers are down, the remaining uplinks handle all the traffic.
However well thought-out our processes and tools may be, mistakes will always happen due to the human factor.Examples include configuration errors, carelessness when working with optical components and disconnecting the wrong cable.
Global LAN has a tried-and-tested team of ten experienced engineers with outstanding qualifications who have been dedicated to keeping the network stable for years.For all the precautions we have taken, nothing can replace experts like these.