- Architecting Cloud Native Applications
- Kamal Arora Erik Farr John Gilbert Piyum Zonooz
- 178字
- 2021-06-24 15:20:57
Providers
For completeness, it is worthwhile to point out that there is a natural bulkhead between cloud providers. All cloud providers will have outages, but the likelihood that they will have simultaneous outages is low. In the context of polyglot cloud, where each component selects the most appropriate provider based on its needs, this simply means that when one provider has an outage then the components running on the other providers will not be affected. Maybe there is an argument to be made that your components should be well diversified across providers, but that is not the objective of polyglot cloud. On the other end of the spectrum, multi-cloud suggests running components redundantly across multiple providers to achieve high availability during an outage. I will wager that maybe 1% of companies have maybe 1% of their workloads which might benefit from all the extra elbow grease that is needed to achieve this type of topology. For the overwhelming majority of us, multi-regional deployments are quite sufficient. Regardless, in cloud-native systems these decisions are made on a component-by-component basis.