
Summary Bullets
• Resilience and diversity can become very expensive
• Architecting a sensibly-priced diverse network is feasible with thought and shopping around
Nearly every single international service provider is involved in cable-build projects in one way or another. For example Verizon has been building an eight-way trans-Pacific mesh network, and expanded its terrestrial mesh into eastern Canada. FT-Orange is increasing capacity in Latin America tenfold, and is involved in new submarine projects such as I-ME-WE (Asia-Europe), EASSy (East Africa) and Lion (linking Madagascar to SAT3/WASC/SAFE). AT&T meanwhile has over 200 Ciena CoreDirectors for control plane–based restoration in its intelligent optical mesh global network. Looking eastwards, several Asian tigers such as NTT Com, China Telecom and China Unicom have all been busy adding extra capacity as well marketing new land-based (e.g., trans-Siberian, Mongolian and through Russia) systems to give more choice and diversity. China Telecom sells the aptly termed ‘Information Silk Route’ as a well chosen brand message for its trans-Siberian network.
Clearly buyers of global, diverse networks should be spreading the net wide in seeking suppliers, because this level of activity and competition results in attractive proposals for a mixture of sub-sea and overland routes to increase resilience. The other options to add resilience open to IT departments are additional equipment line-cards (e.g., 1+1 CPE) representing a fairly costly choice but nonetheless raising the satiability of a network up a another notch, sourcing more than one fiber on a single cable, and as already described sourcing entirely physically separate cables including ones from several carriers. But all of these decisions certainly will add to the final bill. A winning trait inherent in MPLS is connection protection, but potentially adds unwelcome latency during the restoration particularly when confined to a single carrier MPLS platform. Some clients can therefore consider buying unprotected circuits. Finally, you need to look at the risk assessment of application and network down times to your business; and group applications and business functions according to which are the more critical, and only invest in the extra network resilience for supporting applications and mission functions that warrant the higher cost.