Whilst the industry standard is to adopt zonal datacenter redundancy to achieve robustness, Coherence have designed and implemented, with our infrastructure partners, a regional datacenter approach to target 100% availability and reduce latency.
Our two main datacenters are in US-East (Virgina) and EU-West (Dublin), with a third in Asia Pacific (Singapore), to specifically to reduce search index latency for some of our specialist services.
A global load balancer supports latency based routing with a 1 minute failover. In the case of web services, these are a redundant path configured from Zendesk to request services directly to both regional datacenters, to ensure no loss of processing if we were to lose a complete region.
Database redundancy is achieved with regional master-masters, with cross-wired failovers. Application redundancy is achieved with a minimum of two application servers per service type. These auto-scale according to load.
Extensive use of a specialist CDN which can be selectively push expired, provides exceptional global performance and resilience, to support operational services such as WaaS forms.